Key message The Taguchi method and metAFLP analysis were used to optimise barley regenerants towards maximum andminimum levels of tissue culture-induced variation. The subtle effects of symmetric and asymmetric methylationchanges in regenerants were identified. Abstract Plant tissue cultures (PTCs) provide researchers with unique materials that accelerate the development of new breeding cultivars and facilitate studies on off-type regenerants. The emerging variability of regenerants derived from PTCs may have both genetic and epigenetic origins, and may be desirable or degrade the value of regenerated plants. Thus, it is crucial to determine how the PTC variation level can be controlled. The easiest way to manipulate total tissue cultureinduced variation (TTCIV) is to utilise appropriate stress factors and suitable medium components. This study describes the optimisation of in vitro tissue culture-induced variation in plant regenerants derived from barley anther culture, and maximizes and minimizes regenerant variation compared with the source explants. The approach relied on methylation amplified fragment length polymorphism (metAFLP)-derived TTCIV characteristics, which were evaluated in regenerants derived under distinct tissue culture conditions and analysed via Taguchi statistics. The factors that may trigger TTCIV included CuSO 4 , AgNO 3 and the total time spent on the induction medium. The donor plants prepared for regeneration purposes had 5.75% and 2.01% polymorphic metAFLP loci with methylation and sequence changes, respectively. The level of TTCIV (as the sum of all metAFLP characteristics analyzed) identified in optimisation and verification experiments reached 7.51 and 10.46%, respectively. In the trial designed to produce a minimum number of differences between donor and regenerant plants, CuSO 4 and AgNO 3 were more crucial than time, which was not a significant factor. In the trial designed to produce a maximum number of differences between donor and regenerant plants, all factors had comparable impact on variation. The Taguchi method reduced the time required for experimental trials compared with a grid method and suggested that medium modifications were required to control regenerant variation. Finally, the effects of symmetric and asymmetric methylation changes on regenerants were identified using novel aspects of the metAFLP method developed for this analysis.