Purple loosestrife (Lythrum salicaria) is considered one of the worst invasive non-native species of North American wetlands, but its reputation rests on few published, quantitative field studies. This study described the vegetation of two sites (one flooded, one drier) within a heavily invaded wetland in southeastern New York, in order to investigate the common claim that purple loosestrife produces monospecific stands. Native species richness per square meter was 6.8, on average, with mean percent cover ranging from 0.03 to 6.7%, compared to 49% for purple loosestrife. One grass, Phalaris arundinacea, covered on average 31.1% of the area in drier site plots. Twenty-six native taxa were found in a total of 18 1-m² plots. I also conducted an experimental manipulation of the plant community in both sites to demonstrate the role of purple loosestrife in gap colonization. Three 1-m² experimental gaps in each site were recolonized by purple loosestrife at very low density after three years (mean percent cover 1.7), even though recruitment in the drier site was heavy in the first summer. I compared vegetation in the three recolonized gaps, three additional gaps in which purple loosestrife recolonization was prevented, and three unmanipulated control plots. Community structure was not statistically different among these treatments after three years (measured as species richness, Shannon diversity index, or percent cover) because removal plots became dominated by one or two grass or sedge species that replaced purple loosestrife in the community. Recolonized plots contained an average of 6.3 native taxa. Six of the colonizing taxa at each site were not present prior to manipulation. The results show that purple loosestrife stands were not monospecific in this wetland, that other non-native species may dominate if purple loosestrife is removed, and that small gaps may provide regeneration niches for other species.