Seedbank studies often suffer from major methodological inadequacies such as absence of appropriate statistical data analysis and low sampling intensity. Multivariate analysis and computer mapping are innovative ways to treat seedbank data. Computer contour mapping was used to visualize spatial patterns of a population of common lambsquarters at three intervals during a growing season. At one site, high spring seed density of 600 000 seed m-2 was decreased to 18.3% of its original size by July, while at another site, low spring seedbank of common lambsquarters of 25 000 seed m-2 increased to 40 000 seed m-2 by autumn. Seedbank studies usually report results on total seed density or on densities of the most abundant species because of difficulties in analyzing large species matrices using parametric statistics. Multivariate analysis and specifically canonical discriminant analysis (CDA) are well suited for seedbank populations. The seedbanks of six agricultural habitats were demonstrated to be floristically different based on the analysis of the relative abundance of weed species in each site using CDA. Organic soils either under grassland or cultivated had significantly larger total seedbanks than mineral soils. If seedbanks are to be used in predictive population models, quantitative data that are reliable, rapidly obtained with limited resources, and logistically feasible for large sampling protocols are needed. Image analysis may be a potential rapid technique for weed seed recognition of washed soil samples.