To evaluate retention of CPR skills by medical residents (MDs), registered nurses (RNs), we tested single-rescuer CPR skills of 21 MDs, 17 RNs, and 21 laypersons using recording manikin and American Heart Association criteria. All study participants had been trained from 4 to 12 months before testing. No MD or RN and only one layperson performed each step correctly and in proper sequence. If calls for assistance were eliminated, one additional layperson, two MDs, and two RNs performed correctly. There were no significant differences between the MDs and RNs. MDs and RNs did better (p less than .01) in assessment compared to laypersons, but some individuals in each group initiated ventilations and compressions without assessing need. There was no difference in the ability to perform ventilations; all three groups did poorly. MDs and RNs performed compression skills better than laypersons (p less than .01), but all had difficulty with rate and depth of compressions. Moreover, only one-third of the general public demonstrated correct hand placement. Despite more training and experience, MD and RN performance was comparable to layperson performance. These data suggest that improving basic life-support skills could save more lives.