Matrix attachment regions (MARs) are thought to buffer transgenes from the influence of surrounding chromosomal sequences, and therefore to reduce transgene silencing and variation in expression. The statistical properties of more than 400 independent transgenic events produced in Populus, with and without flanking MAR elements from the tobacco root gene RB7, were analysed. The expression of two reporter genes in two poplar clones during three phases of vegetative growth, and the association of T-DNA characteristics with expression, was examined. It was found that MARs did not show a consistent effect on transgene expression levels; they had no effect on the green fluorescent protein (GFP) reporter gene, but reduced expression in the Basta resistance (BAR) reporter gene by 23%. The presence of MARs reduced expression variability within transformant populations, apparently by reducing the number of silenced or weakly expressing events. Transgene expression was highly stable over vegetative growth cycles that spanned 3 years of growth in the glasshouse and field, but MARs showed no association with the strength of correlations in expression over the years. Nonetheless, MARs increased the correlation in expression between a p35S::GFP and prbcS::BAR transgene linked on the same vector, but the effect was small and varied between the years. The presence of MARs had no effect on the transgene copy number, but was positively associated with T-DNA truncations, as well as with the formation of direct over inverted repeats at the same chromosomal locus.