It is important to consider the potential effectiveness of regulations for reducing total harvest levels when developing fishery management plans. A random forest (RF) modelling approach was used to examine how changing per-angler harvest or minimum size limit regulations affected sport fishery harvest in US Atlantic coast recreational fisheries. Harvest limits per angler (i.e. bag limits) were typically high initially and subsequently reduced, whereas almost half of minimum length limits were initially below the length-at-maturity and subsequently increased. Across most fisheries examined, extreme reductions in harvest limits (e.g. from unlimited to catch-and-release) were largely ineffective at limiting total fishery harvest. Increasingly restrictive minimum length limits caused a greater average harvest reduction than per-angler harvest limits. Some regulation changes were associated with higher angling effort and thus increased harvest, which suggests that when effort cannot be constrained, more direct harvest limitations should be considered.