Hey,
just to complement Sislos comment, another way of thinking about it is for MEASUREMENT data (blood pressure, cell counts etc) use T test for TWO (note the T) groups or ANOVA for multiple groups.
CATEGORICAL data use Chi sq. (note the C). Categorical means that it is labeled in categories as in yes/no(smoker/nonsmoker) variables giving counts of people in certain CATEGORY or proportions (percentages) of subjects in a category ....NOTE that come measurement variables can be converted into categorical variables by grouping as an example: Cd4 counts bellow 50, 51-200 and above 200 this would be analyzed by Chi sq.
hope I did not confuse you more with this. let me know if you need a better explanation...