Secondary succession was documented throughout a long-term study of two abandoned hay fields in south-eastern Ontario. In one field ͑Field 1͒, eighteen plots ͑measuring 100 m 2 ͒ were established. Nine of these plots were ploughed to bare ground and nine remained in their original abandoned state. A third set of nine plots were established in a second field ͑Field 2͒, which had been abandoned 5 years earlier. Detailed surveys of the plant flora were conducted twice yearly, insect sweep data were collected bi-monthly, from 1976 Ϫ 1988, 1995 and 1998. Soil and climate data were collected and community patterns analyzed against these variables and field treatment ͑ploughing͒. While the vegetation in all plots began predominantly as a grass mixture, only a few exhibited the typical secondary succession pathway for this region. The remaining plots persist at various stages of succession, with some still in the early stages. Succession stage, as quantified by a succession ratio, varied across these plots largely by soil moisture, and to a lesser degree by field treatment ͑ploughing͒. When grouped by soil moisture, wet plots had lower ratios ͑characterized by later species and trees͒, than dry plots ͑characterized by early or grass species͒. Species richness was highest in plots that were wet, regardless of field treatment. Partial CC analysis using CANOCO of plant species versus field treatment, soil moisture and insect outbreak determined their contribution to the plant variation observed. It is generally accepted that plots over time, regardless of initial composition, will converge to become similar. Convergence of our plots, grouped according to soil moisture, did not occur. Analyses of insect outbreaks against changes in the succession ratio over time provide some evidence that herbivory can influence the likelihood of plots to converge.