This paper presents an approach to estimating the prevalence of cocaine use, based upon a new ratio estimation technique. This method can be applied to random samples of overlapping populations for which no sampling frames exist. When the ratio estimation method is applied to the two study samples (drawn from populations of people using cocaine and people using heroin) the ratio of cocaine users to heroin users (C/H) was 1.55, with a 95% confidence interval of ±0.48. Such estimates should be applied with caution. However, if used with reference to national estimates of about 75 000 heroin users, application of the present estimate suggests that there may be about 116 000 cocaine users in the UK.