In the last few years there has been a growing interest in transport policy concerning behaviour oriented 'soft' measures to reduce private car use. Besides an assessment of the methodological quality of available evaluation results, the present paper focuses on a quantitative, meta-analytical synthesis of this empirical evidence. For these purposes a data set of 141 studies evaluating three types of soft transport policy measures was compiled mainly from already published narrative research reviews. The ability to draw strong causal inferences from the available research evidence is limited by the fact that all the retrieved evaluation studies use weak quasi-experimental designs. At least for one policy measure type our analyses also indicate the presence of a reporting bias. Across all three soft policy measures we found a statistically significant random-effects pooled effect size of 0.15. Translated into the original metric such an effect size indicates an increase in the no-car use proportion from 39% to 46%.