diff --git a/doc/source/release.rst b/doc/source/release.rst
index e9af4ccf50dc4..c5856666ca02a 100644
--- a/doc/source/release.rst
+++ b/doc/source/release.rst
@@ -92,6 +92,8 @@ pandas 0.13
   - Fixed an issue where ``PeriodIndex`` joining with self was returning a new
     instance rather than the same instance (:issue:`4379`); also adds a test
     for this for the other index types
+  - Fixed a bug with all the dtypes being converted to object when using the CSV cparser 
+    with the usecols parameter (:issue: `3192`)
 
 pandas 0.12
 ===========
diff --git a/pandas/io/tests/test_parsers.py b/pandas/io/tests/test_parsers.py
index 730450e373341..d83fbd97b6044 100644
--- a/pandas/io/tests/test_parsers.py
+++ b/pandas/io/tests/test_parsers.py
@@ -2138,6 +2138,28 @@ def test_usecols(self):
         self.assertRaises(ValueError, self.read_csv, StringIO(data),
                           names=['a', 'b'], usecols=[1], header=None)
 
+    def test_usecols_dtypes(self):
+        data = """\
+1,2,3
+4,5,6
+7,8,9
+10,11,12"""
+        result = self.read_csv(StringIO(data), usecols=(0, 1, 2), 
+                               names=('a', 'b', 'c'), 
+                               header=None,
+                               converters={'a': str},
+                               dtype={'b': int, 'c': float},
+                              )                               
+        result2 = self.read_csv(StringIO(data), usecols=(0, 2),
+                               names=('a', 'b', 'c'), 
+                               header=None,
+                               converters={'a': str},
+                               dtype={'b': int, 'c': float},
+                              )
+        self.assertTrue((result.dtypes == [object, np.int, np.float]).all())
+        self.assertTrue((result2.dtypes == [object, np.float]).all())
+
+
     def test_usecols_implicit_index_col(self):
         # #2654
         data = 'a,b,c\n4,apple,bat,5.7\n8,orange,cow,10'
diff --git a/pandas/parser.pyx b/pandas/parser.pyx
index 9bf693f3cb703..36055e681a706 100644
--- a/pandas/parser.pyx
+++ b/pandas/parser.pyx
@@ -869,6 +869,7 @@ cdef class TextReader:
                 if self.has_usecols and not (i in self.usecols or
                                              name in self.usecols):
                     continue
+                nused += 1
 
             conv = self._get_converter(i, name)
 
@@ -907,10 +908,6 @@ cdef class TextReader:
 
             results[i] = col_res
 
-            # number of used column names
-            if i > self.leading_cols:
-                nused += 1
-
         self.parser_start += end - start
 
         return results