We analyze heterogeneity in default effects on retirement saving in a randomized survey experiment. In the Dutch context of universal, mandatory, and high pensions, respondents make realistic choices in which they can maintain the status-quo, suspend, or double pension contributions for one, three, or five years. The aggregate effect sizes of defaults at a three-year suspension and doubling of contributions are similar, increasing the fraction choosing these options by 22%-points. The former most strongly affects those with low incomes and high pension entitlements, while the latter affects patient, non-procrastinating, and especially impulsive individuals. Average savings are influenced in different directions, depending on the alternatives that individuals would have chosen in absence of the default. When setting defaults, it is important to consider variation in counterfactuals.