diff --git a/pyhecdss/__init__.py b/pyhecdss/__init__.py
index b4f1427..2a867d8 100644
--- a/pyhecdss/__init__.py
+++ b/pyhecdss/__init__.py
@@ -1,4 +1,4 @@
 __author__ = """Nicky Sandhu"""
 __email__ = 'psandhu@water.ca.gov'
-__version__ = "0.2.9"
+__version__ = "0.3.0"
 from .pyhecdss import *
diff --git a/pyhecdss/pyhecdss.py b/pyhecdss/pyhecdss.py
index fc1bb2c..4f39255 100755
--- a/pyhecdss/pyhecdss.py
+++ b/pyhecdss/pyhecdss.py
@@ -242,8 +242,11 @@ def m2ihm(self, minute):
     def parse_pathname_epart(self, pathname):
         return pathname.split('/')[1:7][4]
 
-    def _number_between(startDateStr, endDateStr, delta=pd.to_timedelta(1, 'Y')):
-        return (pd.to_datetime(endDateStr)-pd.to_datetime(startDateStr))/delta
+    def _number_between(startDateStr, endDateStr, delta=np.timedelta64(1, 'D')):
+        """
+        This is just a guess at number of values to be read so going over is ok.
+        """
+        return round((pd.to_datetime(endDateStr)-pd.to_datetime(startDateStr))/delta+1)
 
     def _get_timedelta_unit(epart):
         if 'YEAR' in epart:
@@ -441,7 +444,7 @@ def read_its(self, pathname, startDateStr=None, endDateStr=None, guess_vals_per_
         ietime = istime = 0
         # guess how many values to be read based on e part approximation
         ktvals = DSSFile._number_between(startDateStr, endDateStr,
-                                         pd.to_timedelta(1, unit=DSSFile._get_timedelta_unit(epart)))
+                                         np.timedelta64(1, DSSFile._get_timedelta_unit(epart)))
         ktvals = guess_vals_per_block*int(ktvals)
         kdvals = ktvals
         itimes = np.zeros(ktvals, 'i')