Most of the classification algorithms discover flat Fuzzy Classification Rules (FCRs) in ‘If- Then’ form. The knowledge discovered in the form of FCRs allows us to deal with vague, inexact and incomplete premises, however, it ignores exceptions and hierarchies that may exist in data. The simple FCRs enlarge the size of Rule Bases (RBs) with the presence of duplicate clauses that can be removed by arranging the rules in a hierarchical fashion. Moreover, such rules infer incorrect conclusions in the presence of exceptional conditions. This paper proposes the discovery of accurate, interpretable and interesting rules in a novel form named as Fuzzy Hierarchical Censored Classification Rules (FHCCRs) using a Genetic Algorithm approach. The GA design for discovering FHCCRs includes designing of suitable encoding scheme, fitness function and genetic operators. The suggested approach works in three phases: i) fuzzifying a dataset in a pre-processing step, ii) applying a genetic algorithm for discovering FHCCRs and iii) merging FHCCRs into bigger hierarchies in a post-processing step. The proposed approach is applied to five benchmark datasets. It successfully discovers FHCCRs which contain exceptions (also referred as censors) as well as hierarchies. The knowledge discovered in the form of FHCCRs enriches rule bases in respect of interpretability and interestingness.