`Safety first' is a maxim often quoted by those who are wise after the event. The railway industry has recently re-learnt this for itself because attempts to graft a safety argument on to an existing design have delayed many new trains and some infrastructure projects coming into service. Signalling has been a particular problem in both these cases, because of EMC (electromagnetic compatibility) studies between signalling and trains and because of trouble with signalling interfaces with other disciplines. Safety in railways will always need careful application of professional processes; but, if it is planned into a project's development, it need not cause delay. This paper attempts to show how attention to some basic principles can help signalling and communications projects to improve our railways while achieving efficient and economic implementation.