Evaluating the prevalence of lameness within herds of dairy cattle is important for management and certification purposes; however, sampling strategies that could reduce the time taken for an assessment would be valuable. The prevalence of lame and severely lame cows on 224 United Kingdom dairy farms was available for analysis. Presence of more than 1 severely lame cow on a farm was a useful indication of a lameness problem. The vast majority (80%) of the 182 farms that had > or = 1 severely lame cow present had an overall lameness prevalence >25%, whereas only 24% of the 42 farms that had no severely lame cows had an overall prevalence >25%. Information was available on individual milking order through the parlor on the day of the lameness assessment. On 37 farms where cows were housed in a group, lameness prevalence was 11.9% greater in the last third compared with the first third of the milking order. For 36 herds that were larger than 100 cows, sampling a maximum of 100 cows from the middle of the milking order produced an estimate of prevalence within 5% of the true prevalence on 83% of farms. A reasonable sampling strategy may, therefore, be to observe up to 100 cows from the middle of the milking order. Also, presence of severely lame cows at the end of milking may be useful for identifying those farms likely to benefit from further support.