Although left–right items are a standard tool of public opinion research, there is little agreement about the optimal response format. Two disputes can be identified in the literature: (1) whether to provide respondents with a small or large number of answer categories, and (2) whether or not to administer the response scale including a midpoint. This study evaluates the performance of the 101, 11, and 10-point left–right scales, which directly speak to the two disputed aspects of measuring the left–right dimension. Drawing on data from a split ballot multitrait multimethod experiment carried out in a methodological pretest to the German Socio-Economic Panel (SOEP), the analysis shows that the choice of a response format makes a difference in terms of data quality: the 11-point left–right scale produces the highest validity of left–right data closely followed by the 10-point scale. Moreover, an application from electoral research illustrates that the choice of response formats affects substantive interpretations about the nature of the left–right dimension. Since all three scales perform about equally well in terms of reliability and the ease of administration, the findings suggest that the 11-point left–right scale should be used in survey research.