Programs of drug testing welfare recipients are increasingly common in US states and have been considered elsewhere. Though often intensely debated, such programs are complicated to evaluate because their aims are ambiguous-aims like saving money may be in tension with aims like referring people to treatment. We assess such programs using a proportionality approach, which requires that for ethical acceptability a practice must be reasonably likely to meet its aims, sufficiently important in purpose as to outweigh harms incurred, and lower in costs than feasible alternatives. In the light of empirical findings, we argue that the programs fail the three requirements. Pursuing recreational drug users is not important in the light of costs incurred, while dependent users who may require referral are usually identifiable without testing and typically need a broader approach than one focussing on drugs. Drug testing of welfare recipients is therefore not ethically acceptable policy.