All vascular plants were inventoried in 64 SE Norwegian landscape ponds and their adjacent margins; two separate species lists were made for each sampling unit. Individual study sites varied considerably in species richness, 1–20 and 13–81 for ponds and pond margins, respectively. A total of 56 explanatory variables were recorded for each pond and adjacent margin.Data on species composition and species richness were analysed separately for ponds and pond margins. Vegetation gradients were found by parallel use of the two ordination methods DCA (Detrended Canonical Analysis) and GNMDS (Global Non-metric Multidimensional Scaling). Interpretation of ordination axes was made by using correlation analyses, GLM (generalised linear modelling) and by geostatistical analyses of spatial structure. Patterns of species richness were analysed by correlation analyses and GLM. The first DCA and GNMDS ordination axes both for ponds and pond margins were strongly correlated and the main gradient was related to geographical variables (such as UTM northing, altitude, distance to forest), pond age and water chemical variables. Water depth and soil depth also explained some of the variation in species composition in ponds and pond margins, respectively. The second DCA and GNMDS axes were different for the two data sets. The second gradient for pond margins was also related to geography and water chemistry in addition to some of the anthropological variables, whereas the second gradient for ponds was harder to find. Correlations and GLM analyses of species richness revealed that mainly water chemistry, in addition to periodical pond drainage and liming were significant predictors of pond species richness, whereas area, if the pond had recently been expanded, some water chemical and geographical variables were significant for the species richness of ponds margins.Different structuring processes which may contribute to explaining variation in species composition and richness are discussed. The analysis of spatial structure of species composition and explanatory variables showed that the ecological data were weakly spatially structured over large range of scales and particular patterns were hard to find. Together with the generally weak explanatory power of the selected variables, this indicates high importance of apparent randomness in this ecosystem, and, notably, that the ponds and their adjacent margins represent islands in the agricultural landscape that accumulate species more or less individualistically.