Shared space is an approach to street design which minimises demarcations between vehicles and pedestrians. It has become particularly influential in the UK, where a comprehensive study of shared space schemes has informed recently published national guidance to local highway authorities. This paper critically examines the claim made in the guidance that it is 'evidence based'. Primary research reported in the paper examines one of the sites in the 'official study' in Ashford, Kent, in greater depth, using video observation and a street survey of pedestrians. The findings show that most pedestrians diverted away from their desire lines, gave way to vehicles in most cases and felt safer under the original road layout. This evidence, and the analysis of the 'official study', cast doubt on some aspects of the methodology and its interpretation in the national guidance. The authors conclude that some of the claims made on behalf of shared space have overstated the available evidence, and that caution is needed in implementing shared space schemes, particularly in environments of high traffic flows.