This paper discusses the possibility of managing search direction in genetic algorithm crossover and mutation operators using “Depository of Weak Individuals’ Fitness Values”. In order to control crossover and mutation, it was decided to observe the individuals that take part in those operators are observed. All offsprings that are produced during crossover operator run are compared with the individuals that are their predecessors – worse individual fitness value is kept in the “Depository of Weak Individuals’ Fitness Values” during all generations. Mutation operator consists of several steps – individual’s mutation will be repeated until better offspring is found. If better solution is found, then the individual selected for mutation will be kept during all generations in the “Depository of Weak Individuals Fitness Values”. To observe the influence of provided modifications on algorithm convergence, experiments using constrained multimodal functions are performed.