Age-stratified data on toxoplasma seroprevalence in pregnant women in Stockholm, Sweden for the years 1969, 1979 1987 provide the basis for an analysis of temporal patterns of Toxoplasma gondii infection, and estimation of the risk maternal toxoplasmosis, in this population. A catalytic infection model, in which the rate or force of infection is assumed to be a function of time (and not, as is more usual, age), was employed to describe the observed changes in levels toxoplasma seropositivity. A range of simple incidence functions (up to 3 parameters) were fitted using a method maximum likelihood. The data were significantly better described by a linear or an exponential decay in the rate of infection through time compared with a constant level. More complex incidence functions gave no better data description. Thus, whilst there is strong evidence for declining incidence in Stockholm over the past 4–5 decades, the data do not allow discrimination between different possibilities for the nature of this decline. Based on these modelling results, best estimates of the force of infection in 1987 acting on susceptible women are within the range 0 to 0·0045/susceptible/year (95% confidence limits), yielding a possible risk of maternal toxoplasmosis of between 0 and 2·7 cases/1000 pregnancies. These values are shown to be significantly lower than estimates based upon an assumption of temporal stability in toxoplasma incidence, which may be of practical significance to public health policy.