Many analyses for single-case designs (SCDs)--including nearly all the effect size indicators-- currently assume no trend in the data. Regression and multilevel models allow for trend, but usually test only linear trend and have no principled way of knowing if higher order trends should be represented in the model. This paper shows how Generalized Additive Models (GAMs) can be used to inform this aspect of the analysis. It is well known in traditional interrupted time series analysis that correctly modeling trend in the data is essential to obtaining an accurate effect size estimate. Nonlinearities can be inherent in the data (e.g., weight loss that slows over time resulting in a quadratic trend) or can result from interactions of the treatment with time (e.g., after a stable baseline, a treatment slowly becomes more effective as more sessions occur). Modeling trend is difficult in SCDs because they rarely have the large number of observations on a case over time to allow use of traditional methods such as ARIMA modeling. Analyses of SCDs by ordinary regression or multilevel models can address trend, but require the researcher to impose the particular functional form, something they can only intuit from visual inspection of the graph. GAMs are a semi-parametric regression model that allows the data to inform the required functional form. The authors show how they can be applied to SCDs. Given the early stage of this research, the authors propose the use of GAMs as a sensitivity analysis for whether trend might affect the main conclusion about treatment effectiveness. They believe, We believe, however, they have potential to become a primary data analytic tool for SCD data. This research should inform SCD researchers about the need to take linear or nonlinear trend into account. With further study, it may also be that GAMs may prove to be a primary analytic method for use in SCDs.