Macro-algae represent an ideal resource of third generation biofuels, but their use necessitates a refinement of commonly used anaerobic digestion processes. In a previous study, contrasting mixes of dairy slurry and the macro-alga Ulva lactuca were anaerobically digested in mesophilic continuously stirred tank reactors for 40 weeks. Higher proportions of U. lactuca in the feedstock led to inhibited digestion and rapid accumulation of volatile fatty acids, requiring a reduced organic loading rate. In this study, 16S pyrosequencing was employed to characterise the microbial communities of both the weakest (R1) and strongest (R6) performing reactors from the previous work as they developed over a 39 and 27-week period respectively. Comparing the reactor communities revealed clear differences in tax-onomy, predicted metabolic orientation and mechanisms of inhibition, while constrained canonical analysis (CCA) showed ammonia and biogas yield to be the strongest factors differentiating the two reactor communities. Significant biomarker taxa and predicted metabolic activities were identified for viable and failing anaerobic digestion of U. lactuca. Acetoclastic methanogens were inhibited early in R1 operation, followed by a gradual decline of hydrogenotrophic methanogens. Near-total loss of methanogens led to an accumulation of acetic acid that reduced performance of R1, while a slow decline in biogas yield in R6 could be attributed to inhibition of acetogenic rather than methanogenic activity. The improved performance of R6 is likely to have been as a result of the large Methanosarcina population, which enabled rapid removal of acetic acid, providing favourable conditions for substrate degradation.