1. Marine Protected Areas (MPAs) are currently proliferating in the Mediterranean Sea. The assessment of their role in the protection and recovery of overexploited fish populations, however, is not yet completely supported by an adequate body of scientific evidence. This is in part because the number of studies that unambiguously assess changes induced by protection is limited.2. Sampling to detect and measure the expected effects of protection on commercial fish (e.g. increase in mean abundance and size) is made difficult by intrinsic natural spatial and temporal variability. Ideally, temporal replication before and after the establishment of MPAs is necessary for estimating the average conditions in time. Spatial replication of control locations, on the other hand, is essential to prevent confusion in interpreting a difference between protected and unprotected locations. Until now, the basic requirements for appropriate experimental designs in MPAs were seldom satisfied, especially in the Mediterranean region.3. In the present paper, an ideal experimental design aimed at overcoming many of the confounding influences that could result in misinterpretations of the results is considered, with a discussion of implications of applying less and less appropriate designs because of pragmatic constraints. Experimental design should thus be adapted to each individual MPA under study on the basis of the specific constraints that researchers face (e.g. 'before' (i.e. pre-designation) data are lacking; there is a single or more than one protected locations).4. The issues related to MPAs, chiefly in the Mediterranean region, need to be considered to enable the identification of unambiguous hypotheses prior to sampling, based on clear logical structures. The importance of appropriate sampling and the subsequent interpretation of data would progressively reduce the degree of uncertainty of environmental analyses about the effects of MPAs on fish, with important implications for their management and further proliferation.