Existing evidence on the effectiveness of automatic enrolment is focused on large employers. We compare pension savings of employees working for small employers who were pseudo-randomly affected by automatic enrolment with those working for small employers who, at the same moment in time, were not. We find that automatic enrolment substantially increased workplace pension participation among those working for small employers by around 45 percentage points to reach 70% of targeted employees – with most, but not all, brought in at relatively low rates of pension saving. Despite this large increase, this leaves pension participation substantially below the very high levels (around 90%) seen under automatic enrolment among the largest employers in the United Kingdom and the United States. We find evidence that this lower participation rate is not explained either by differences in the observed characteristics of workers between smaller and larger employers, or by differences in the pension contributions offered by employers.