Context Data collected from producers in south-eastern Australia found that the survival of twin-born lambs decreased by 3.5% for each extra 100 ewes in the mob at lambing. Increasing stocking rate by 1 ewe/ha decreased lamb survival by a further 0.7%. These survey findings suggest that lamb survival could be improved by optimising the allocation of ewes to mobs and paddocks at lambing. Aim The present paper reports two experiments. Experiment 1 tested the hypotheses that (1) the survival of twin-born lambs would be greater when ewes lamb in smaller mobs and at lower stocking rates, and (2) the effects of mob size and stocking rate would be greater in Merinos than in non-Merino breeds. Experiment 2 tested the hypothesis that the survival of twin-born Merino lambs would be greater at lower mob sizes when ewes lambed at stocking rates <4 ewes/ha. Methods Experiment 1 investigated a 2 × 2 factorial combination of mob size (high or low) and stocking rate (high or low) on the survival of twin-born Merino and non-Merino lambs at 70 on-farm research sites across southern Australia. Experiment 2 investigated the effect of high or low mob size on the survival of twin-born Merino lambs when ewes lambed at stocking rates of <4 twin ewes/ha at 15 on-farm research sites. In both experiments, adult twin-bearing ewes were randomly allocated into a treatment and lambing paddock on Day 140 from the start of joining at each farm. Lamb survival in each mob was calculated based on lamb losses between pregnancy scanning and lamb marking. Key results In both experiments, the effect of mob size on lamb survival was found to be linear, with survival of twin-born lambs decreasing by between 1.9% and 2.5% per additional 100 ewes in the mob at lambing, regardless of breed (P < 0.001). In Experiment 1, there was no effect of stocking rate or mob size by stocking rate on lamb survival. Conclusions The present research demonstrated that reducing mob size but not stocking rate will improve the survival of twin-born lambs to marking for extensive enterprises in Australia where ewes lamb at stocking rates of up to 12 ewes/ha. Implications These findings will contribute to guidelines for optimising ewe nutrition and resource allocation to improve lamb survival.