Colonization rates of mussels (Mytilus trossulus and M. edulis) were measured on natural substrata in tidepools and on emergent rock in recently ice-scoured and non-scoured regions of a rocky shore near Halifax, Nova Scotia, Canada. The relative importance of initial settlement/colonization, compared to subsequent dispersal and mortality, in determining the distribution and abundance of mussels was examined by comparing patterns and rates of mussel colonization at sampling intervals of days to months over a 17 mo period. Lrss than 4 ":, of mussels which colonized the quadrats sampled at short (2 to 7 d) intervals were settling larvae (2 mm, too large to be dispersed by byssal drifting, suggesting they \Yere redistributed by wave dislodgment and deposition. At both short (2 to 7 d) and long (5 to 16 mo) sarnpllng ~ntervdls, colon~sts were most abundant in ice-scoured tidepools and least abundant on ice-scoured emergent rock, probably reflecting d~fferences In the macrobenthic assemblaqe. the sllhc;tratl~m fnr co!oniza!ic~. 1 :: addi:ion, thr-luny L~I I I I dbunaance oi colonists was linearly related to the cumulativct short term abundance during all but one of the intervals. Therefore, our results ~ndicate that, over time scales up to 16 mo, patterns of initial colonization by settlers and larger post-larval mussels were inore important than post-colonizati.on mortality and dispersal in determining pattt=rns of distribution and abundance of n~ussels on this shore