The effectiveness of marine protected areas depends largely on whether people comply with the rules. We quantified temporal changes in benthic composition, reef fish biomass, and fishing effort among marine park zones (including no-take areas) to assess levels of compliance following the 2005 rezoning of the government-controlled Karimunjawa National Park (KNP), Indonesia. Four years after the rezoning awareness of fishing regulations was high amongst local fishers, ranging from 79.5±7.9 (SE) % for spatial restrictions to 97.7±1.2% for bans on the use of poisons. Despite this high awareness and strong compliance with gear restrictions, compliance with spatial restrictions was weak. In the four years following the rezoning reef fish biomass declined across all zones within KNP, with >50% reduction within the no-take Core and Protection Zones. These declines were primarily driven by decreases in the biomass of groups targeted by local fishers; planktivores, herbivores, piscivores, and invertivores. These declines in fish biomass were not driven by changes in habitat quality; coral cover increased in all zones, possibly as a result of a shift in fishing gears from those which can damage reefs (i.e., nets) to those which cause little direct damage (i.e., handlines and spears). Direct observations of fishing activities in 2009 revealed there was limited variation in fishing effort between zones in which fishing was allowed or prohibited. The apparent willingness of the KNP communities to comply with gear restrictions, but not spatial restrictions is difficult to explain and highlights the complexities of the social and economic dynamics that influence the ecological success of marine protected areas. Clearly the increased and high awareness of fishery restrictions following the rezoning is a positive step. The challenge now is to understand and foster the conditions that may facilitate compliance with spatial restrictions within KNP and marine parks worldwide.