Background: The ADAGIO study investigated whether rasagiline has disease-modifying effects in Parkinson's disease. Rasagiline 1 mg per day, but not 2 mg per day, was shown to be efficacious in the primary analysis. Here, we report additional secondary and post-hoc analyses of the ADAGIO study. Methods: ADAGIO was a placebo-controlled, double-blind, multicentre, delayed-start study, in which 1176 patients with untreated early Parkinson's disease were randomly assigned to receive rasagiline 1 mg or 2 mg per day for 72 weeks (early-start groups) or placebo for 36 weeks followed by rasagiline 1 mg or 2 mg per day for 36 weeks (delayed-start groups). We assessed the need for additional antiparkinsonian therapy and changes in non-motor experiences of daily living and fatigue scales (prespecified outcomes) and changes in unified Parkinson's disease rating scale (UPDRS) scores and subscores in placebo and active groups (post-hoc outcomes). The ADAGIO study is registered with ClinicalTrials.gov, number NCT00256204. Findings: The need for additional antiparkinsonian therapy was reduced with rasagiline 1 mg (25 of 288 [9%] patients) and 2 mg (26 of 293 [9%]) versus placebo (108 of 593 [18%]; odds ratio for 1 mg rasagiline vs placebo 0·41, 95% CI 0·25-0·65, p=0·0002; 2 mg rasagiline vs placebo 0·41, 0·26-0·64, p=0·0001). At week 36, both doses significantly improved UPDRS motor subscores compared with placebo (1 mg rasagiline mean difference -1·88 [SE 0·35]; 2 mg rasagiline -2·18 [0·35]; both p<0·0001) and activities of daily living subscores (ADL; 1 mg rasagiline -0·86 [0·18]; 2 mg rasagiline -0·88 [0·18]; both p<0·0001), and 1 mg rasagiline significantly improved UPDRS mentation subscore (-0·22 [0·08]; p=0·004). At week 72, the only significant difference between early-start and delayed-start groups was for ADL subscore with the 1 mg dose (-0·62 [0·29]; p=0·035). When assessed for the effect on non-motor symptoms at week 36, both doses showed benefits on the Parkinson fatigue scale versus placebo (1 mg rasagiline mean difference -0·14 [SE 0·05], p=0·0032; 2 mg rasagiline -0·19 [0·05], p<0·0001), and the 1 mg dose showed benefits on the scale for non-motor experiences of daily living compared with placebo (mean difference -0·33 [0·17]; p=0·049). The rate of progression of total UPDRS score for patients in the placebo group was 4·3 points [SE 0·3] over 36 weeks, with extrapolation to about 6 units per year. In the placebo group, patients with the lowest quartile of baseline UPDRS scores (≤14; n=160) progressed more slowly than did those with highest scores (>25·5; n=145; mean difference -3·46 [SE 0·77]; p<0·0001). Interpretation: These findings show that rasagiline delayed the need for symptomatic antiparkinsonian drugs and emphasise the contribution of the UPDRS ADL in the response of the rasagiline 1 mg per day early-start versus delayed-start group. The rate of UPDRS deterioration was less than was anticipated from previous studies and correlated with baseline severity. Understanding of the pattern of UPDRS deterioration is essential to assess disease modification. Funding: Teva Pharmaceutical Industries and H Lundbeck A/S.