One of the most important features of a database, beside collecting and storing information, is its ability to extract and produce the required data within the quickest possible time. In many cases, during the modelling and design stage, the designers and analysts concentrate their efforts on the building the structure, leaving the data access time aside. The article presents a relational database time-based tuning system with the use of the fuzzy logic. Database in-time tuning is based mainly on database analysis (both on logical and physical level), SQL response time analysis and then creating more efficient indexes upon those analyses. The tuning is usually done by the people involved directly in applications needing efficiency improvement, i.e. the programmers, analysts or database administrators. The tuning is done based on data obtained from the system with the help of programs supplied along with the database management system. The experts, however, tend to use the expressions like 'huge amount of data', 'long response time', which strongly binds the optimisation with the knowledge, intuition and the expert's reality perception. Thus, it seems reasonable to apply the fuzzy logic to the in-time database structure tuning, in order to, for example, shorten the response time. The detailed information resulting from the database system analysis, together with the experts knowledge defined as sets and rules of fuzzy logic's concluding, thus closer to the natural language, can be a powerful tool in database tuning.