rounding errors between SQL and .NET - why helps to use decimal type instead of float

I came across a situation where when the database datatype is float, and we read the data into a .NET datatable, the automatic conversion is from float to double. For some reason when this happens there are occasionally cases of rounding errors. The way to overcome this (other than changing the float columns to decimal(18, 6) or some other decimal spec, is to access the float data using some CONVERT(decimal(18, 6) col_name_here) so that the .NET data table has decimal type in the column in question.