The prevalence of pathological gambling in the general population is considered a relevant parameter when prevention and treatment issues are discussed. These estimates generally result from large general population surveys. For practical and economic reasons large surveys cannot rely on specifically trained highly qualified experts, but have to be performed by more or less well-trained professional interviewers. The latter can only achieve their related to low prevalent phenomena satisfactory if they can use highly reliable and valid assessment tools. Since the prevalence of pathological gambling in many countries lies in the one-percent range, meaningful results are only conceivable if the psychometric properties of these instruments are truly excellent. Scientists developing screening instruments, validating them and/or using them commonly claim that the psychometric properties of their instruments are good to excellent – based on empirical results that actually suggest something quite different – and commonly they get away without dissent. The presentation will use simple calculations based on generally accepted formulas, to demonstrate and quantify the problems arising when the sensitivity and/or specificity of the applied tools is suboptimal. Subsequently, the question will be discussed whether the application of common screening tools and questionnaires in general population surveys to assess low prevalent phenomena is justified, if instruments are far from perfect.