[en] Changes of fuel element design and modifications of the operational conditions have to be tested in experiments and pilot projects for nuclear safety. Experimental design is an useful statistical method minimizing costs and risks for this procedure. The main problem of our work was to investigate the connection between failure rate of fuel elements, sample size, confidence interval, and error probability. Using the statistic model of the binomial distribution appropriate relations were derived and discussed. A stepwise procedure based on a modified sequential analysis according to Wald was developed as a strategy of introduction for modifications of the fuel element design and of the operational conditions. (author)