Selfing can evolve if the transmission advantage of selfers outweighs the negative effects of inbreeding depression. It has been hypothesised that on the long term, selfing lineages are an evolutionary dead end, in part due to genetic degradation resembling that of Muller's ratchet in asexual lineages. There is a lack of empirical evidence for costs of selfing due to genetic degradation in recently evolved selfers. We tested whether such costs are apparent in recently established selfing populations of the generally outcrossing species Arabidopsis lyrata. Specifically, we compared selfing and outcrossing populations in their growth performance, and for traits that play a putative role in defence against herbivores. In line with our expectations, selfing populations had reduced germination rates, growth however was similar to outcrossing populations. Plants from selfing popu lations showed no consistent reduction in herbivore defence traits, and were equally palatable to caterpillars of the moth Mamestra brassicae. There were also no differences between outcrossers and selfers in phenotypic plasticity for putative defence traits and palatability after induction by herbivores. Overall, we interpret our results as showing some evidence for persistent costs of selfing due to drift or inbreeding load in terms of reduced seedling establishment, but providing no support for the hypothesis that selfing populations should be more susceptible to generalist herbivores, or rely more on induced defence.