Five split-ballot experiments, plus replications, were carried out in several national surveys to compare the effects of offering or omitting a middle alternative in forced-choice attitude questions. Explicitly offering a middle position significantly increases the size of that category, but tends not to otherwise affect univariate distributions. The relation of intensity to the middle position is somewhat greater on Offered forms than on Omitted forms (less intense respondents being more affected by question form than those who feel more strongly), but in general form does not alter the relationship between an item and a number of other respondent characteristics. Finally, in one instance there is evidence that form can change the conclusion about whether two attitude items are related, but the results are of uncertain reliability.