It is crucial for valid statistical inference to make appropriate assumptions about the data generating process, which is why specific assumptions have to be formulated very carefully and non-parametric tests are preferable in case of any doubt about the underlying population. Although the two-tailed sign test is probably the most popular non-parametric test for location problems, the practitioner is confronted with serious drawbacks, such as the handling of ties and an essential loss of data order information, which result from the assumption of crisp categories. In this paper, we present a two-tailed sign test with fuzzy formulated categories that overcomes these shortcomings and allows to incorporate expert knowledge and user’s priorities into the test procedure. Moreover, the proposed test is neat in theory and practice and avoids disadvantages that are often associated with fuzzy tests (such as a fuzzy test decision). We perform an extensive real case study related to the body mass index and the detection of associated considerable health risks. The results of the case study clearly indicate that embedding of fuzzy categories improves the performance of the two-tailed sign test.