NatCen has just published its report on the effectiveness of the much hailed Association of British Bookmakers' Code of Practice. The Code, people may remember, was hurriedly put together when it looked as if there was a serious possibility that betting company profits might be affected by significant tightening up of the rules about playing Fixed Odds Betting Terminals. At the time most commentators could see that this action by ABB was too little to have much effect. But it does seem to have been highly effective in the sense that it has headed Government off from taking any further action. The research just published appears to show, as could have been predicted, that the Code of Practice had a negligible effect. The Code was introduced by five of the biggest betting company chains at the beginning of March 2014. Its introduction was delayed for one month in a control region: the West Midlands. That provided a small opportunity to test the Code's initial impact. The Code required betting shops to offer all customers playing their machines the opportunity to engage in 'pre-commitment' by setting their own personal loss or time limits. One test of the Code's effectiveness was therefore how many players actually set such limits in practice. In fact the proportions of playing sessions where players did so was tiny: in less than 0.1% of sessions were loss limits set, with an even lower percent where time limits were set (itself interesting because there is some evidence that setting time limits is the more effective). The percentages were slightly higher in the few weeks after the Code was first introduced, but they quickly settled down to those lower percentages. Since so few people took the opportunity to set limits, it is not surprising that the comparison of the West Midlands with the rest of the country for the one month before the Code was introduced in the West Midlands showed no significant difference in the length of betting sessions or the amounts of money players lost. The proportion of sessions where players were alerted, as required by the Code, because they had inserted £250 or more during the session or had been playing for 30 minutes or more were substantially higher but still quite small. Mandatory money alerts were activated during 3.0-3.5% of sessions. Mandatory time alerts were activated during approximately 6.5% of sessions. It is important to bear in mind that this piece of research has its limitations. If there had been a really serious wish to carry out a controlled study, then a bigger control group should have been used (not just the West Midlands) and there should have been a longer delay before the Code was introduced there. But the pressure to get on with introducing the Code across the whole country was such that the one-month delay for the West Midlands was all that could be agreed. It was also the case that the betting data available to researchers was confined to those who held loyalty cards with the betting companies. Although there is known to be a concentration of gambling problems amongst loyalty cardholders, that does represent a shortcoming. The other thing to say is that the Code also included sections about improved staff training, including training in recognising problem gambling; those aspects of the Code were outside the remit of the research just published.