Objective: This randomized cross-over trial aimed to determine whether listening to music during cardiopulmonary resuscitation (CPR) training increased the proportion of lay persons performing chest compressions correctly.Methods: 130 lay volunteers (81 male) were recruited opportunistically at a UK university.Those aged less than 18 years, or with CPR training within three months, and health professionals were excluded.Volunteers performed three sequences of one minute of continuous chest compressions on a manikin accompanied by no music (NM); choruses of Nellie the Elephant (Nellie); and Thats the Way (I like it) (TTW) in a pre-randomized order.Results: Median and inter-quartile range (IQR) for compression rate were NM 110 (93 to 119); Nellie 105 (98 to 107); TTW 109 (103 to 110): between-intervention differences were significant for Nellie vs.NM and Nellie vs.TTW (p<0.001) but not NM vs.TTW (p=0.055).A compression rate between 95 and 105 was achieved with NM, Nellie,and TTW for 15/130 (12%), 42/130 (32%) and 12/130 (9%) attempts respectively: differences were significant for Nellie vs.NM and Nellie vs.TTW (p<0.0001) but not NM vs.TTW (p=0.55).Relative risk for a compression rate between 95 and 105 was 2.8 for Nellie vs.NM (95%CI 1.66-4.80), 0.8 for TTW vs.NM (95%CI 0.40-1.62), and 3.5 for Nellie vs.TTW (95%CI 1.97-6.33).The median proportion of compressions which were too shallow was higher when participants listened to Nellie vs.NM (56 vs.47%, P=0.022).Conclusions: Listening to Nellie the Elephant significantly increased the proportion of lay persons delivering compression rates at close to 100 per minute, but also increased the proportion of compressions which were too shallow.Since current resuscitation guidelines give equal emphasis to correct compression rate and depth the authors recommend that using Nellie the Elephant as a learning aid during CPR training be discontinued.