Automatic vs. manual plotting
When Prism performs nonlinear regression, it automatically superimposes the curves on the graph.
If you need to create additional graphs, or change which curve is plotted on which graph, keep in mind that from Prism's point of view, a curve generated by nonlinear regression is simply a data set. You can add curves to a graph or remove curves from a graph on the 'Data on graph' tab of the Format Graph dialog.
If you ask Prism to compare two models, only the selected model is graphed. If you want to graph both models, you will need to analyze the data twice, once with each model.
With Prism 5 and 6, it is Ok to stretch an axis to a log scale, a choice on top of the Format Axis dialog. With prior versions, this didn't work well. The curve was defined as a series of line segments, and these would be far from equally spaced on a log axis. Prism 5 and 6 are smarter, so the curve will look smooth on either a linear or logarithmic axis.
Let's say that the X values of your data range from 0 to 100, and you fit a curve with nonlinear regression. Now you want to create a graph showing only the data with X values between 0 and 10. The curve will look ragged. The problem is that Prism defines the entire curve using 150 line segments. In this example, you only plot 10% of the range, so the curve will have only 15 line segments.
To fix this problem, go to the Range tab of nonlinear regression and choose to define the curve with more line segments. In this example, increasing the value tenfold, to 1500, would be about right.