A new distinction between two kinds of state constraints is introduced and proved important for solving the Ramiication Problem in Reasoning about Actions. Steady constraints never, not even for an instant, cease to being in force. As such they give rise to truly instantaneous indirect eeects of actions. Stabilizing state constraints, on the other hand, may be suspended for a short period of time after an action has occurred. Indirect eeects deriving from these constraints materialize with a short causal lag. Carelessly mixing these two types of indirect eeects is shown to producing anomalous models. On the basis of the existing theory of causal relationships we illustrate how to deal with the distinction appropriately.