Thanks for your email. I don't have much bootstrapping experience. I attach my SAS code as below. Could you please help me take a look and see if I put it in the right way? My data is 500 MRI rate, 97 of them are zeros, one is 100, about 20 more than 50, the rest of them is between 0 and 50. The average is 14.6.   data test;   set impw;   diff = mri_per_100_episodes - average; run; proc surveyselect data = test out=outboot   seed=0   method=urs   samprate=1   outhits   rep=500; run; proc univariate data = outboot;   var diff;   output out=result pctlpts=2.5, 97.5 pctlpre=ci; run;   Thanks Nancy     Date: Thu, 16 Apr 2009 10:03:52 -0500 From: Warren Schlechte <Warren.Schlechte@TPWD.STATE.TX.US> Subject: Re: How to get Confidence intervals in my data Unless your bootstrapping ignored some structure of the data -- which it may have -- or it was computed incorrectly -- which it could have been -- the fact that your CI is "too wide" is likely a true reflection of your data. Maybe a better explanation of why it's too wide and not ideal for you could get others to help. Warren Schlechte