Basal and canopy cover of sown and unsown species in swards sown with six species mixtures were assessed monthly from autumn 2010 to spring 2014, to test the hypotheses that (a) sowing an alternative pasture base, or (b) increasing the complexity of the sown mix, improves persistence and reduces weed ingress in temperate summer-dry dairy pastures. Treatments comprised either perennial ryegrass (Lolium perenne L.) infected with AR1 endophyte or tall fescue (Schedonorus arundinaceus (Schreb.) Dumort.) infected with Max P endophyte to which were added either a legume (‘standard’, 2-species mix), a legume and two forage herbs (‘herbs’, 4-species), or three legumes, two forage herbs and two grasses (‘complex’, 8-species). In the first year, basal and canopy cover of sown species were higher in perennial ryegrass- than tall fescue-based swards, and basal cover of sown species was higher and the percentage bare ground lower in the ‘standard’ (50%) than ‘herbs’ and ‘complex’ swards (42%). By the final year, basal cover of sown species (25%), unsown species (28%), and percentage bare ground (47%) were similar in all six treatments. Although establishment was greater in perennial ryegrass than tall fescue-based swards and in the ‘standard’ than in the ‘herbs’ and ‘complex’ mixtures, the loss of sown species in these treatments was greater. The decline in basal cover of sown species was –27% in the standard treatment, –16% averaged over the ‘complex’ and ‘herbs’ treatment, –24% in perennial ryegrass-based swards and –15% in tall fescue-based swards. The results are contrary to both hypotheses with respect to weed ingress. However, support was provided for the hypotheses in the greater persistence (smaller decline over time in basal cover) in tall fescue than ryegrass-based swards, and ‘herbs’ and ‘complex’ than ‘standard’ mixtures.