Governments are increasingly using algorithms to automate decision-making procedures and generate administrative decisions that previously required human oversight and judgement. However, the black-boxing of algorithmic procedures and the automation of street-level discretion lead to concerns about administrative organizations’ ability to balance the need for universal and predictable rules with the need for fairness, proportionality and accountability in individual administrative decisions. In this contribution, we apply classic principles of good administration to automated decision-making and propose additional safeguards to mitigate the risks of automation for citizens. A principle-based approach to the use of algorithms is needed to design in reviewability of procedures and give citizens means to hold administrative organizations accountable for their decisions. However, for principles of good digital administration to have an actual effect, fundamental challenges remain in the institutionalization of these principles and in dealing with the epistemological opacity of algorithms.