Measurement of chest velocity has been proposed as an alternative method to identify responder leaning during cardiopulmonary resuscitation (CPR). Leaning is defined in terms of force, but no study has tested the utility of chest velocity in the presence of force measurements that directly measure leaning. We analyzed 1004 out-of-hospital cardiac arrest (OHCA) files collected with Q-CPR monitors in the Portland, Oregon, USA metro region from 2006 to 2017. Records contained accelerometry and force signals. For each chest compression, the following metrics were computed: minimum force at the end of the compression (Frelease), compression depth, compression rate, maximum chest velocity during recoil (vrecoil) and maximum rate of change in force during chest release (ʋrelease). A compression was classified as having leaning if Frelease was greater than 2.5 kg-f. The ability of vrecoil and ʋrelease to predict Frelease was estimated with generalized linear models, and their ability to identify leaning with logistic regression. The data set contained over 1.5 million chest compressions, 21% compliant with 2015 rate and depth guidelines for CPR (the G2015 population). Leaning was uncommon generally (12%), and less common in G2015 compliant compressions (5%). Leaning and Frelease decreased with both vrecoil and ʋrelease but with extensive overlap. Neither vrecoil nor ʋrelease, alone or in combination with chest compression rate and depth, reliably predicted leaning or Frelease. Leaning cannot be reliably identified from vrecoil or ʋrelease, alone or in combination with currently recommended chest compression metrics in out-of-hospital CPR.