Adherence to HIV treatment regimes is a core element to viral suppression. Yet measurement of adherence is complex. Although adherence levels are good predictors of outcome, they do not always provide full explanations of observed variations in responses. This study was set up to examine the complexity of adherence measurement and to examine rates of adherence in the presence of complex measurement. A total of 502 consecutive attenders at HIV clinics in the UK (80.5% response rate) provided detailed measurement on adherence in the preceding 7 days, setting out dose adherence, as well as measures of timing and dietary conditions. In addition, a range of psychological, demographic and relationship data were gathered to understand predictors of full and partial adherence. Although 79.1% reported dose adherence in the previous 7 days, 42.8% had not taken the dose at the correct time, and 27.2% had not taken the dose under the correct circumstances. Using a more complex composite measure of full adherence, rates reduced from 79.1% to 41.5%. Comparisons of those deemed fully adherent, partially adherent and non-adherent were carried out. Those that were fully adherent were significantly more likely to be older (F=7.8, p<0.001), UK born (F=6.8, p=0.03), code ethnicity as white (F=5.3, p=0.07), record higher quality of life (chi(2)=8.7, p=0.01), lower psychological symptoms (chi(2)=15.2, p=0.001) and lower global distress symptoms (chi(2)=6.9, p=0.03). There were no differences according to education, behavioural and attitudinal variables (disclosure, stable relationship, STI diagnosed, number of sexual partners, unprotected sex, optimism or treatment switching). Fully adherent groups were significantly more likely to be in agreement with their doctor on treatment initiation (chi(2)=6.2, p=0.045), satisfied with the amount of involvement in the decision-making process (chi(2)=7.3, p=.026), their wishes were considered (chi(2)=12.5, p=0.002) and had monitoring of their condition (chi(2)=7.1, p=0.028). Multivariate analysis showed that variables which contributed significantly at a 5% criterion level to complex adherence were physical symptoms (OR=0.56, p=0.05), psychological symptoms (OR=2.37, p<0.001), age (OR=0.96, p=0.02), education (OR=0.54, p=0.03), having more than one sexual partner (OR=0.46, p=0.03), having risky sex (OR=4.30, p=0.002) and being optimistic about treatments (OR=0.42, p=0.01). The softer markers of adherence are not usually measured in follow up and may account for variations in treatment responses. The complexity of adherence needs to be understood and addressed to maximise treatment efficacy.