Oil-pollution monitoring at sea through beach bird surveying would undoubtedly benefit from a further standardisation of methods, enhancing the efficiency of data collection. In order to come up with useful recommendations, we evaluated various approaches of beached bird collection at the Belgian coast during seven winters (1993-1999). Data received in a passive way by one major rehabilitation centre were compared to the results of targeted beach surveys carried out at different scales by trained ornithologists: ‘weekly’ surveys - with a mean interval of 9 days - restricted to a fixed 16.7 km beach stretch, ‘monthly’ surveys over the entire coastline (62.1 km) and an annual ‘international’ survey in Belgium over the same distance at the end of February. Data collected through Belgian rehabilitation centres concern injured, living birds collected in a non-systematic way. Oil rates derived from these centres appear to be strongly biased to oiled auks and inshore bird species, and are hence of little use in assessing the extent of oil pollution at sea. The major asset of rehabilitation centres in terms of data collection seems to be their continuous warning function for events of mass mortality. Weekly surveys on a representative and large enough section rendered reliable data on oil rates, estimates of total number of bird victims, representation of various taxonomic groups and species-richness and were most sensitive in detecting events quickly (wrecks, oil-slicks, severe winter mortality,…). Monthly surveys gave comparable results, although they overlooked some important beaching events and demonstrated slightly higher oil rates, probably due to the higher chance to miss short-lasting wrecks of auks. Since the monthly surveys in Belgium were carried out by a network of volunteers and were spread over a larger beach section, they should be considered as best performing. Single ‘international beached bird surveys’ in February gave reliable data on total victim number (once the mean ratio between numbers in various months is known) and oil rate (provided a sufficiently large sample can be collected), but failed in tracking events. It is a particularly attractive approach because of its long tradition, resulting in invaluable long-term databases, and the uniformity in which these surveys are organised on a large scale. The minimal distance for a monthly survey amounts to 25-30 km (40-50% of Belgian coastline) up to 40 km (65%) in order to attain sound figures for oil rate and species-richness respectively. These distances are primarily determined by the number of bird corpses that may be collected and are hence a function of beaching intensity and corpse detection rate.