Soft-bottom beds of the blue mussel (Mytilus edulis) are of high ecological importance in intertidal and subtidal habitats. They create habitat, shelter and food for other organisms, and play a dominant role in energy flow and nutrient cycling. Intertidal beds are much better studied than subtidal beds. Though it is often assumed that subtidal mussel beds resemble their intertidal counterparts, major differences in factors driving recruitment, growth and survival can be expected. The aim of our study was to estimate survival chances of the mussel beds in the subtidal parts of the Dutch Wadden Sea in relation to environmental variables, and to compare the results with those obtained previously from the intertidal areas. We used data from a long-term annual survey, resulting in a survival analysis of 365 individual subtidal mussel beds. The average life span of subtidal mussel beds, once they have survived their first winter, was estimated at 2.3 years. This is lower than what was found in the intertidal (3.4 years) in a previous study. However, the survival of subtidal mussel beds in less-saline areas is comparable to survival of intertidal mussel beds, whereas survival of subtidal mussel beds in more-saline areas is significantly lower. The strong, significant effect of salinity is most likely an indication of an effect from starfish predation, since starfish (Asterias rubens) are virtually absent from the intertidal and their abundance is strongly reduced in the subtidal at lower salinities. Furthermore, the survival of individual beds is positively correlated with their size. This may be a direct effect of the bed size itself, or also an indirect effect of environmental factors that can affect the size of newly settled beds. A secondary aim was to compare two methods, based on different types of field data: 1) empirical point data and 2) estimated bed contours based on the point data and additional sources of information. Both methods give similar results. Advantages and disadvantages of both methods are discussed.