diff --git a/pandas/plotting/_core.py b/pandas/plotting/_core.py
index 81f5b5cb0f74c..78c7082c69b6b 100644
--- a/pandas/plotting/_core.py
+++ b/pandas/plotting/_core.py
@@ -5,19 +5,16 @@
 from pandas.core.dtypes.common import is_integer, is_list_like
 from pandas.core.dtypes.generic import ABCDataFrame, ABCSeries
 
-import pandas
 from pandas.core.base import PandasObject
 from pandas.core.generic import _shared_doc_kwargs, _shared_docs
 
-# Automatically registering converters was deprecated in 0.21, but
-# the deprecation warning wasn't showing until 0.24
-# This block will be eventually removed, but it's not clear when
-if pandas.get_option('plotting.matplotlib.register_converters'):
-    try:
-        from .misc import register
-        register(explicit=False)
-    except ImportError:
-        pass
+# Trigger matplotlib import, which implicitly registers our
+# converts. Implicit registration is deprecated, and when enforced
+# we can lazily import matplotlib.
+try:
+    import pandas.plotting._matplotlib  # noqa
+except ImportError:
+    pass
 
 df_kind = """- 'scatter' : scatter plot
         - 'hexbin' : hexbin plot"""
diff --git a/pandas/plotting/_matplotlib/__init__.py b/pandas/plotting/_matplotlib/__init__.py
index 5cfb6843db9ed..1b775d03349d0 100644
--- a/pandas/plotting/_matplotlib/__init__.py
+++ b/pandas/plotting/_matplotlib/__init__.py
@@ -1,3 +1,5 @@
+from pandas._config import get_option
+
 from pandas.plotting._matplotlib.boxplot import (
     BoxPlot, boxplot, boxplot_frame, boxplot_frame_groupby)
 from pandas.plotting._matplotlib.converter import deregister, register
@@ -11,6 +13,10 @@
 from pandas.plotting._matplotlib.timeseries import tsplot
 from pandas.plotting._matplotlib.tools import table
 
+if get_option("plotting.matplotlib.register_converters"):
+    register(explicit=False)
+
+
 __all__ = ['LinePlot', 'BarPlot', 'BarhPlot', 'HistPlot', 'BoxPlot', 'KdePlot',
            'AreaPlot', 'PiePlot', 'ScatterPlot', 'HexBinPlot', 'hist_series',
            'hist_frame', 'boxplot', 'boxplot_frame', 'boxplot_frame_groupby',
diff --git a/pandas/plotting/_matplotlib/boxplot.py b/pandas/plotting/_matplotlib/boxplot.py
index b8a7da5270fc0..f8bc531e3c344 100644
--- a/pandas/plotting/_matplotlib/boxplot.py
+++ b/pandas/plotting/_matplotlib/boxplot.py
@@ -1,7 +1,6 @@
 from collections import namedtuple
 import warnings
 
-from matplotlib import pyplot as plt
 from matplotlib.artist import setp
 import numpy as np
 
@@ -11,6 +10,7 @@
 import pandas as pd
 
 from pandas.io.formats.printing import pprint_thing
+from pandas.plotting._matplotlib import converter
 from pandas.plotting._matplotlib.core import LinePlot, MPLPlot
 from pandas.plotting._matplotlib.style import _get_standard_colors
 from pandas.plotting._matplotlib.tools import _flatten, _subplots
@@ -215,6 +215,7 @@ def boxplot(data, column=None, by=None, ax=None, fontsize=None,
             rot=0, grid=True, figsize=None, layout=None, return_type=None,
             **kwds):
 
+    import matplotlib.pyplot as plt
     # validate return_type:
     if return_type not in BoxPlot._valid_return_types:
         raise ValueError("return_type must be {'axes', 'dict', 'both'}")
@@ -296,6 +297,8 @@ def plot_group(keys, values, ax):
 def boxplot_frame(self, column=None, by=None, ax=None, fontsize=None, rot=0,
                   grid=True, figsize=None, layout=None,
                   return_type=None, **kwds):
+    import matplotlib.pyplot as plt
+    converter._WARN = False  # no warning for pandas plots
     ax = boxplot(self, column=column, by=by, ax=ax, fontsize=fontsize,
                  grid=grid, rot=rot, figsize=figsize, layout=layout,
                  return_type=return_type, **kwds)
@@ -306,6 +309,7 @@ def boxplot_frame(self, column=None, by=None, ax=None, fontsize=None, rot=0,
 def boxplot_frame_groupby(grouped, subplots=True, column=None, fontsize=None,
                           rot=0, grid=True, ax=None, figsize=None,
                           layout=None, sharex=False, sharey=True, **kwds):
+    converter._WARN = False  # no warning for pandas plots
     if subplots is True:
         naxes = len(grouped)
         fig, axes = _subplots(naxes=naxes, squeeze=False,
diff --git a/pandas/plotting/_matplotlib/core.py b/pandas/plotting/_matplotlib/core.py
index a7049afee80b0..5fb4d201223bd 100644
--- a/pandas/plotting/_matplotlib/core.py
+++ b/pandas/plotting/_matplotlib/core.py
@@ -2,7 +2,6 @@
 from typing import Optional  # noqa
 import warnings
 
-import matplotlib.pyplot as plt
 import numpy as np
 
 from pandas._config import get_option
@@ -61,6 +60,8 @@ def __init__(self, data, kind=None, by=None, subplots=False, sharex=None,
                  secondary_y=False, colormap=None,
                  table=False, layout=None, **kwds):
 
+        import matplotlib.pyplot as plt
+        converter._WARN = False  # no warning for pandas plots
         self.data = data
         self.by = by
 
@@ -103,7 +104,7 @@ def __init__(self, data, kind=None, by=None, subplots=False, sharex=None,
             self.rot = self._default_rot
 
         if grid is None:
-            grid = False if secondary_y else self.plt.rcParams['axes.grid']
+            grid = False if secondary_y else plt.rcParams['axes.grid']
 
         self.grid = grid
         self.legend = legend
@@ -618,6 +619,8 @@ def _get_ax(self, i):
 
     @classmethod
     def get_default_ax(cls, ax):
+        import matplotlib.pyplot as plt
+
         if ax is None and len(plt.get_fignums()) > 0:
             with plt.rc_context():
                 ax = plt.gca()
diff --git a/pandas/plotting/_matplotlib/hist.py b/pandas/plotting/_matplotlib/hist.py
index 585c407e33311..d34c0cb6a3889 100644
--- a/pandas/plotting/_matplotlib/hist.py
+++ b/pandas/plotting/_matplotlib/hist.py
@@ -1,6 +1,5 @@
 import warnings
 
-import matplotlib.pyplot as plt
 import numpy as np
 
 from pandas.core.dtypes.common import is_integer, is_list_like
@@ -10,6 +9,7 @@
 import pandas.core.common as com
 
 from pandas.io.formats.printing import pprint_thing
+from pandas.plotting._matplotlib import converter
 from pandas.plotting._matplotlib.core import LinePlot, MPLPlot
 from pandas.plotting._matplotlib.tools import (
     _flatten, _set_ticks_props, _subplots)
@@ -203,6 +203,7 @@ def _grouped_hist(data, column=None, by=None, ax=None, bins=50, figsize=None,
     def plot_group(group, ax):
         ax.hist(group.dropna().values, bins=bins, **kwargs)
 
+    converter._WARN = False  # no warning for pandas plots
     xrot = xrot or rot
 
     fig, axes = _grouped_plot(plot_group, data, column=column,
@@ -220,6 +221,7 @@ def plot_group(group, ax):
 def hist_series(self, by=None, ax=None, grid=True, xlabelsize=None,
                 xrot=None, ylabelsize=None, yrot=None, figsize=None,
                 bins=10, **kwds):
+    import matplotlib.pyplot as plt
     if by is None:
         if kwds.get('layout', None) is not None:
             raise ValueError("The 'layout' keyword is not supported when "
@@ -261,6 +263,7 @@ def hist_series(self, by=None, ax=None, grid=True, xlabelsize=None,
 def hist_frame(data, column=None, by=None, grid=True, xlabelsize=None,
                xrot=None, ylabelsize=None, yrot=None, ax=None, sharex=False,
                sharey=False, figsize=None, layout=None, bins=10, **kwds):
+    converter._WARN = False  # no warning for pandas plots
     if by is not None:
         axes = _grouped_hist(data, column=column, by=by, ax=ax, grid=grid,
                              figsize=figsize, sharex=sharex, sharey=sharey,
diff --git a/pandas/plotting/_matplotlib/misc.py b/pandas/plotting/_matplotlib/misc.py
index dacc9ef04f819..663a3c5153fac 100644
--- a/pandas/plotting/_matplotlib/misc.py
+++ b/pandas/plotting/_matplotlib/misc.py
@@ -2,7 +2,6 @@
 
 import matplotlib.lines as mlines
 import matplotlib.patches as patches
-import matplotlib.pyplot as plt
 import numpy as np
 
 from pandas.core.dtypes.missing import notna
@@ -105,6 +104,7 @@ def _get_marker_compat(marker):
 
 
 def radviz(frame, class_column, ax=None, color=None, colormap=None, **kwds):
+    import matplotlib.pyplot as plt
 
     def normalize(series):
         a = min(series)
@@ -169,6 +169,7 @@ def normalize(series):
 
 def andrews_curves(frame, class_column, ax=None, samples=200, color=None,
                    colormap=None, **kwds):
+    import matplotlib.pyplot as plt
 
     def function(amplitudes):
         def f(t):
@@ -224,6 +225,7 @@ def f(t):
 
 def bootstrap_plot(series, fig=None, size=50, samples=500, **kwds):
 
+    import matplotlib.pyplot as plt
     # random.sample(ndarray, int) fails on python 3.3, sigh
     data = list(series.values)
     samplings = [random.sample(data, size) for _ in range(samples)]
@@ -270,6 +272,7 @@ def parallel_coordinates(frame, class_column, cols=None, ax=None, color=None,
                          use_columns=False, xticks=None, colormap=None,
                          axvlines=True, axvlines_kwds=None, sort_labels=False,
                          **kwds):
+    import matplotlib.pyplot as plt
     if axvlines_kwds is None:
         axvlines_kwds = {'linewidth': 1, 'color': 'black'}
 
@@ -336,6 +339,7 @@ def parallel_coordinates(frame, class_column, cols=None, ax=None, color=None,
 
 def lag_plot(series, lag=1, ax=None, **kwds):
     # workaround because `c='b'` is hardcoded in matplotlibs scatter method
+    import matplotlib.pyplot as plt
     kwds.setdefault('c', plt.rcParams['patch.facecolor'])
 
     data = series.values
@@ -350,6 +354,8 @@ def lag_plot(series, lag=1, ax=None, **kwds):
 
 
 def autocorrelation_plot(series, ax=None, **kwds):
+    import matplotlib.pyplot as plt
+
     n = len(series)
     data = np.asarray(series)
     if ax is None:
diff --git a/pandas/plotting/_matplotlib/style.py b/pandas/plotting/_matplotlib/style.py
index 80a15942a2867..8c9e3ea330dd3 100644
--- a/pandas/plotting/_matplotlib/style.py
+++ b/pandas/plotting/_matplotlib/style.py
@@ -3,7 +3,6 @@
 
 import matplotlib.cm as cm
 import matplotlib.colors
-import matplotlib.pyplot as plt
 import numpy as np
 
 from pandas.core.dtypes.common import is_list_like
@@ -13,6 +12,7 @@
 
 def _get_standard_colors(num_colors=None, colormap=None, color_type='default',
                          color=None):
+    import matplotlib.pyplot as plt
     if color is None and colormap is not None:
         if isinstance(colormap, str):
             cmap = colormap
diff --git a/pandas/plotting/_matplotlib/timeseries.py b/pandas/plotting/_matplotlib/timeseries.py
index 30038b599a386..e36ffed10d94f 100644
--- a/pandas/plotting/_matplotlib/timeseries.py
+++ b/pandas/plotting/_matplotlib/timeseries.py
@@ -3,8 +3,6 @@
 import functools
 import warnings
 
-from matplotlib import pylab
-import matplotlib.pyplot as plt
 import numpy as np
 
 from pandas._libs.tslibs.frequencies import (
@@ -42,6 +40,7 @@ def tsplot(series, plotf, ax=None, **kwargs):
     .. deprecated:: 0.23.0
        Use Series.plot() instead
     """
+    import matplotlib.pyplot as plt
     warnings.warn("'tsplot' is deprecated and will be removed in a "
                   "future version. Please use Series.plot() instead.",
                   FutureWarning, stacklevel=2)
@@ -323,6 +322,7 @@ def format_dateaxis(subplot, freq, index):
     default, changing the limits of the x axis will intelligently change
     the positions of the ticks.
     """
+    from matplotlib import pylab
 
     # handle index specific formatting
     # Note: DatetimeIndex does not use this
diff --git a/pandas/plotting/_matplotlib/tools.py b/pandas/plotting/_matplotlib/tools.py
index f6393fc76892f..e491cfc3309a0 100644
--- a/pandas/plotting/_matplotlib/tools.py
+++ b/pandas/plotting/_matplotlib/tools.py
@@ -2,7 +2,6 @@
 from math import ceil
 import warnings
 
-import matplotlib.pyplot as plt
 import matplotlib.table
 import matplotlib.ticker as ticker
 import numpy as np
@@ -168,6 +167,7 @@ def _subplots(naxes=None, sharex=False, sharey=False, squeeze=True,
     # Four polar axes
     plt.subplots(2, 2, subplot_kw=dict(polar=True))
     """
+    import matplotlib.pyplot as plt
     if subplot_kw is None:
         subplot_kw = {}
 
@@ -345,6 +345,7 @@ def _get_xlim(lines):
 
 def _set_ticks_props(axes, xlabelsize=None, xrot=None,
                      ylabelsize=None, yrot=None):
+    import matplotlib.pyplot as plt
     for ax in _flatten(axes):
         if xlabelsize is not None:
             plt.setp(ax.get_xticklabels(), fontsize=xlabelsize)
diff --git a/pandas/tests/plotting/test_converter.py b/pandas/tests/plotting/test_converter.py
index 39cd48ff35f96..92d207e46b7ab 100644
--- a/pandas/tests/plotting/test_converter.py
+++ b/pandas/tests/plotting/test_converter.py
@@ -12,11 +12,30 @@
 from pandas import Index, Period, Series, Timestamp, date_range
 import pandas.util.testing as tm
 
+from pandas.plotting import (
+    deregister_matplotlib_converters, register_matplotlib_converters)
 from pandas.tseries.offsets import Day, Micro, Milli, Second
 
-converter = pytest.importorskip('pandas.plotting._converter')
-from pandas.plotting import (deregister_matplotlib_converters,  # isort:skip
-                             register_matplotlib_converters)
+try:
+    from pandas.plotting._matplotlib import converter
+except ImportError:
+    # try / except, rather than skip, to avoid internal refactoring
+    # causing an improprer skip
+    pass
+
+pytest.importorskip('matplotlib.pyplot')
+
+
+def test_initial_warning():
+    code = (
+        "import pandas as pd; import matplotlib.pyplot as plt; "
+        "s = pd.Series(1, pd.date_range('2000', periods=12)); "
+        "fig, ax = plt.subplots(); "
+        "ax.plot(s.index, s.values)"
+    )
+    call = [sys.executable, '-c', code]
+    out = subprocess.check_output(call, stderr=subprocess.STDOUT).decode()
+    assert 'Using an implicitly' in out
 
 
 def test_timtetonum_accepts_unicode():
diff --git a/pandas/tests/plotting/test_datetimelike.py b/pandas/tests/plotting/test_datetimelike.py
index 10743ca95e29e..c3d824389aa4d 100644
--- a/pandas/tests/plotting/test_datetimelike.py
+++ b/pandas/tests/plotting/test_datetimelike.py
@@ -374,7 +374,6 @@ def test_axis_limits(self):
         def _test(ax):
             xlim = ax.get_xlim()
             ax.set_xlim(xlim[0] - 5, xlim[1] + 10)
-            ax.get_figure().canvas.draw()
             result = ax.get_xlim()
             assert result[0] == xlim[0] - 5
             assert result[1] == xlim[1] + 10
@@ -383,7 +382,6 @@ def _test(ax):
             expected = (Period('1/1/2000', ax.freq),
                         Period('4/1/2000', ax.freq))
             ax.set_xlim('1/1/2000', '4/1/2000')
-            ax.get_figure().canvas.draw()
             result = ax.get_xlim()
             assert int(result[0]) == expected[0].ordinal
             assert int(result[1]) == expected[1].ordinal
@@ -392,7 +390,6 @@ def _test(ax):
             expected = (Period('1/1/2000', ax.freq),
                         Period('4/1/2000', ax.freq))
             ax.set_xlim(datetime(2000, 1, 1), datetime(2000, 4, 1))
-            ax.get_figure().canvas.draw()
             result = ax.get_xlim()
             assert int(result[0]) == expected[0].ordinal
             assert int(result[1]) == expected[1].ordinal
@@ -429,12 +426,7 @@ def test_get_finder(self):
     def test_finder_daily(self):
         day_lst = [10, 40, 252, 400, 950, 2750, 10000]
 
-        if self.mpl_ge_3_0_0 or not self.mpl_ge_2_2_3:
-            xpl1 = xpl2 = [Period('1999-1-1', freq='B').ordinal] * len(day_lst)
-        else:  # 2.2.3, 2.2.4
-            xpl1 = [7565, 7564, 7553, 7546, 7518, 7428, 7066]
-            xpl2 = [7566, 7564, 7554, 7546, 7519, 7429, 7066]
-
+        xpl1 = xpl2 = [Period('1999-1-1', freq='B').ordinal] * len(day_lst)
         rs1 = []
         rs2 = []
         for i, n in enumerate(day_lst):
@@ -457,12 +449,7 @@ def test_finder_daily(self):
     def test_finder_quarterly(self):
         yrs = [3.5, 11]
 
-        if self.mpl_ge_3_0_0 or not self.mpl_ge_2_2_3:
-            xpl1 = xpl2 = [Period('1988Q1').ordinal] * len(yrs)
-        else:  # 2.2.3, 2.2.4
-            xpl1 = [68, 68]
-            xpl2 = [72, 68]
-
+        xpl1 = xpl2 = [Period('1988Q1').ordinal] * len(yrs)
         rs1 = []
         rs2 = []
         for i, n in enumerate(yrs):
@@ -485,12 +472,7 @@ def test_finder_quarterly(self):
     def test_finder_monthly(self):
         yrs = [1.15, 2.5, 4, 11]
 
-        if self.mpl_ge_3_0_0 or not self.mpl_ge_2_2_3:
-            xpl1 = xpl2 = [Period('Jan 1988').ordinal] * len(yrs)
-        else:  # 2.2.3, 2.2.4
-            xpl1 = [216, 216, 204, 204]
-            xpl2 = [216, 216, 216, 204]
-
+        xpl1 = xpl2 = [Period('Jan 1988').ordinal] * len(yrs)
         rs1 = []
         rs2 = []
         for i, n in enumerate(yrs):
@@ -521,11 +503,7 @@ def test_finder_monthly_long(self):
 
     @pytest.mark.slow
     def test_finder_annual(self):
-        if self.mpl_ge_3_0_0 or not self.mpl_ge_2_2_3:
-            xp = [1987, 1988, 1990, 1990, 1995, 2020, 2070, 2170]
-        else:  # 2.2.3, 2.2.4
-            xp = [1986, 1986, 1990, 1990, 1995, 2020, 1970, 1970]
-
+        xp = [1987, 1988, 1990, 1990, 1995, 2020, 2070, 2170]
         xp = [Period(x, freq='A').ordinal for x in xp]
         rs = []
         for i, nyears in enumerate([5, 10, 19, 49, 99, 199, 599, 1001]):
@@ -1093,7 +1071,6 @@ def test_time(self):
         df.plot(ax=ax)
 
         # verify tick labels
-        fig.canvas.draw()
         ticks = ax.get_xticks()
         labels = ax.get_xticklabels()
         for t, l in zip(ticks, labels):
@@ -1120,7 +1097,6 @@ def test_time_change_xlim(self):
         df.plot(ax=ax)
 
         # verify tick labels
-        fig.canvas.draw()
         ticks = ax.get_xticks()
         labels = ax.get_xticklabels()
         for t, l in zip(ticks, labels):
@@ -1138,7 +1114,6 @@ def test_time_change_xlim(self):
         ax.set_xlim('1:30', '5:00')
 
         # check tick labels again
-        fig.canvas.draw()
         ticks = ax.get_xticks()
         labels = ax.get_xticklabels()
         for t, l in zip(ticks, labels):
@@ -1165,7 +1140,6 @@ def test_time_musec(self):
         ax = df.plot(ax=ax)
 
         # verify tick labels
-        fig.canvas.draw()
         ticks = ax.get_xticks()
         labels = ax.get_xticklabels()
         for t, l in zip(ticks, labels):
@@ -1432,7 +1406,7 @@ def test_format_timedelta_ticks_narrow(self):
         df = DataFrame(np.random.randn(len(rng), 3), rng)
         fig, ax = self.plt.subplots()
         df.plot(fontsize=2, ax=ax)
-        fig.canvas.draw()
+        self.plt.draw()
         labels = ax.get_xticklabels()
 
         result_labels = [x.get_text() for x in labels]
@@ -1456,7 +1430,7 @@ def test_format_timedelta_ticks_wide(self):
         df = DataFrame(np.random.randn(len(rng), 3), rng)
         fig, ax = self.plt.subplots()
         ax = df.plot(fontsize=2, ax=ax)
-        fig.canvas.draw()
+        self.plt.draw()
         labels = ax.get_xticklabels()
 
         result_labels = [x.get_text() for x in labels]
@@ -1529,7 +1503,7 @@ def test_matplotlib_scatter_datetime64(self):
         df["time"] = date_range("2018-01-01", periods=10, freq="D")
         fig, ax = self.plt.subplots()
         ax.scatter(x="time", y="y", data=df)
-        fig.canvas.draw()
+        self.plt.draw()
         label = ax.get_xticklabels()[0]
         if self.mpl_ge_3_0_0:
             expected = "2017-12-08"
diff --git a/pandas/util/_test_decorators.py b/pandas/util/_test_decorators.py
index 4cc316ffdd7ab..e81a51e99f5a3 100644
--- a/pandas/util/_test_decorators.py
+++ b/pandas/util/_test_decorators.py
@@ -75,7 +75,7 @@ def safe_import(mod_name, min_version=None):
 def _skip_if_no_mpl():
     mod = safe_import("matplotlib")
     if mod:
-        mod.use("Agg", warn=False)
+        mod.use("Agg", warn=True)
     else:
         return True