Background: Studies on rewording word problems can be grouped into two main groups: situational rewording, in which the situation denoted by the text is described more richly, and conceptual rewording, in which the underlying semantic relations are highlighted. Aims: Our aims are to define and distinguish these two kinds of rewording and to test empirically their relative effectiveness in two different studies. Sample: In the first study, 79 third graders, 64 fourth graders and 65 fifth graders took part; the sample for Study 2 was similar. Method: In Study 1, children were asked to solve both easy and difficult two-step change problems in three different versions: standard, situational and conceptual rewording. In Study 2, three different versions of the situational version were compared: one with only temporal elaborations, one with only causal elaborations and a "complete" version combining both elaborations. Results: In Study 1, conceptually reworded problems elicited the best results, especially among younger children and for difficult two-step problems. Neither in Study 1 nor in Study 2 did the situationally reworded problems yield better performance than standard items. Conclusion: Only conceptual rewording has proved to be useful for improving children's performance, especially among younger children and for difficult problems. The lack of impact of situational rewording cannot be explained in terms of the length of the resulting text.