Background: In March 2008, the D:A:D study published results showing an increased risk of myocardial infarction (MI) in those on abacavir (ABC). We describe changes in use of ABC since publication, and reinvestigate this association with subsequent follow-up. Methodology: D:A:D participants were followed from study entry until the first of: MI, death, 1st Feb 2013 or end of follow up. Associations between 10-year Framingham cardiovascular disease (CVD) risk (low [<10%], moderate [10-20%], high [>20%] or unknown [U/K]) and likelihood of initiating or discontinuing ABC were assessed with multivariable logistic/Poisson regression. Poisson regression also assessed the association between current ABC use (with 6-month lag to allow for recent discontinuation) and MI risk, adjusting for use of other ART drugs and confounders. Analyses were performed separately pre- and post-March 2008. Results: Use of ABC increased from 10% in 2000 to 20% in 2008, before stabilising at 18-19% (Figure). Increases in use pre-March 2008, and subsequent decreases, were greatest in those at moderate and high CVD risk. Post-March 2008, those on ABC at moderate/high CVD risk were more likely to discontinue ABC than those at low/unknown CVD risk, regardless of VL (<1000 cps/ml: relative rate (RR) 1.49 [95% CI 1.34-1.65]; >1000 cps/ml: 1.23 [1.02-1.48]); no such associations were seen pre-March 2008. There was some evidence that ART-naive people at moderate/high CVD risk post-March 2008 were less likely to initiate ABC than those at low/unknown CVD risk (odds ratio 0.74 [0.48-1.13]). By 1st Feb 2013, 941 MI events occurred in 367,559 person years (PYRS) (rate 0.26 [0.24-0.27]/100 PYRS). Rate of MI was 0.47 [0.42-0.52] for current ABC use and 0.21 [0.19-0.22] otherwise. Current ABC use was associated with a 98% increase in MI rate (RR 1.98 [1.72-2.29]), with no difference in the pre- (1.97 [1.68-2.33]; 672 events, 210,250 PYRS) or post- (1.97 [1.43-2.72]; 269 events, 157,309 PYRS) March 2008 periods (interaction p=0.74). Results were unchanged after further adjusting for factors potentially on the causal pathway (pre-March 2008: 1.88 [1.55-2.28]; post-March 2008: 2.03[1.46-2.84]), including renal function, dyslipidaemia and hypertension. Conclusions: Despite channelling of ABC away from those at higher CVD risk since 2008, we continue to observe an association between ABC use and MI risk. Whilst confounding cannot be ruled out in any cohort study, this argues against channelling bias as an explanation for our findings.