We carried out a somewhat similar analysis based on the results of the first British Gambling Prevalence Survey (BGPS). Questions designed to assess the 10 DSM-IV criteria were asked of all participants who acknowledged engaging in gambling in the previous 12 months (n = 5550). We also found a strong first factor (accounting for 40% of total variance) with all but one of the individual criteria loading 0.4 or above. Again, the exceptional item was chasing, which loaded only 0.31 on the first general factor. In the light of Gerstein et al.'s (2003) paper I have looked in more detail at how individual criteria related to total DSM scores in our data. The results look similar. Comparing those whose total score was 1–2, 3–4 and 5 or above (numbers are too small to break the latter group down further, the percentage meeting the chasing losses criterion varies little (51%, 54%, 60%), whereas the gradient for committing crime to finance gambling is steep, particularly at the higher end (<1%, 13%, 50%).