Since its discovery in Dunkeld (Scotland) at the beginning of the 20th century, hybrid larch (HL) (Larix x eurolepis) raised a strong interest among foresters and, later on, breeders. Dozens of hybridization orchards have been established in several European countries. All of them include selected genotypes of European Larch (EL) (Larix decidua Mill.) and Japanese larch (JL) (Larix kaempferi (Lamb.) Carr.) but they differ in the mother species (EL, JL, both), the number of clones used on both maternal and paternal sides (from one to several dozens) and in the planting design. However, the products of these seed orchards are composed of a mixture of hybrid and pure species seeds, the latter resulting from intraspecific crossings and/or self-pollination. Molecular markers show highly variable hybrid percentages among orchards but also, for a given orchard, among years. In addition to the number of mother clones, the proportion of hybrid seeds depends on several factors, e.g. clonal contributions to pollen and seed-cone production and flowering overlapping that are related to genetic factors, climatic conditions and seed orchard management (flower stimulation, supplemental mass pollination). Considering the extent of hybrid rate fluctuations (between 10% and 90%) it is important to study the effect of species purity on stand productivity and quality. The consequences are also discussed in the context of seed orchard testing. Our data have been mainly collected in two "twin" trials managed by Irstea that aimed at comparing seed lots from six European hybridization orchards (64 seedlings per plot x 4 blocks for each variety). They were planted in 1995 in two sites, at mid-elevation (800 and 1000 m). The first one is subjected to an oceanic climate favourable to JL whereas the second one, more continental, is more suitable to EL. Bedrock is granite in site 1 and limestone in site 2. Taxa were first identified (100-120 trees per variety) by INRA using markers based on cytoplasmic DNA. It allowed us to compare the performances of hybrids and pure species and to study the evolution of HL percentage over time and thinnings. Hybrid varieties were: FH 201, Halle, Esbeek, Maglehem (in both sites) as well as FP 237 and Vaals in site 2. Seed lots were produced in seed orchards with various designs (one or several maternal clones, seeds collected from EL or JL). Hybrid proportion varied tremendously in the tested populations. It was low in FP 237 and Halle (12-26%), intermediate in Maglehem (43%) and high in FH 201, Vaals and Esbeek (84-96%). Taxa did not vary for survival and adaptation because no sanitary problem occurred in the trials after age 10-14 when taxa were identified. However, there was some evidence that pure species were more affected by early mortality than hybrids in some varieties. For each variety and in both sites, hybrids had a better growth in height and diameter. Still, huge differences were found among varieties. At age 6, superiority of hybrids over pure species varied from 30% (Halle) to 500% (FP 237) for stem volume. Those differences can be partly explained by orchard composition, e.g. the species used as mother and the number of maternal clones. In addition, the last measurements showed that hybrids continue to grow faster than pure species after thinning. Regarding stem form, hybrids were generally less slender than EL or JL. On average, they were also somewhat less straight but the differences were rarely significant. Simulations of thinning show that the plots will become richer in hybrids if the trees are selected for vigour. Still, the final percentage of hybrids will remain far below 100% for the varieties characterised by a slight difference of growth between hybrids and pure species and/or low hybrid rate at planting. On the other hand, hybrid percentage is expected to remain more or less stable if stem form is the character taken into account. As form and growth are not correlated, thinning based on both traits will result in an intermediate enrichment in hybrids, at a level depending on the weight assigned to each character. From a silvicultural point of view, the impact of hybrid rate depends on both forest owner strategy and the variety. For "stem form" objective, Forest Reproductive Material (FRM) richness in hybrids does not matter much (whatever the variety among those studied) but it would be nonsense to use expensive hybrid larch seedlings for that sole purpose. On the contrary, hybrid percentage is of major importance for timber production and its impact varies according to the variety. When hybrids do not differ much from pure species, the loss of productivity should be moderate even with medium hybrid rates. In varieties where hybrids grow much faster than pure species, the risk of poor timber production is high. Yet, the consequences are probably negligible if the hybrid rate is higher than 60-70% because most of the pure species would be eliminated after the first thinning that has no commercial value in standard silviculture. However, such FRM are probably not suitable for biomass production. When testing seed orchard progenies, tree status (hybrid or not) is generally unknown, which may cause misinterpretation of the results. Firstly, there can be a large difference between observed performances and the variety “intrinsic” value, i.e. hybrid performances. This occurs in populations composed of few vigorous hybrids mixed with a lot of self-pollinated individuals, which is the case of FP 237 in our trials. However, it is hard to decide what the “actual" value of a variety is because the year-to-year variations of FRM hybrid rate are generally unknown. Secondly, even when tree status is known, comparing the performances of hybrids produced in different orchards may be inaccurate when the hybrid trials grow older. Indeed, competition among trees occurs earlier and is more intense in homogeneous varieties that might be at a disadvantage by comparison with varieties with a bimodal distribution. Considering the importance of hybrid purity, the huge variations across seed and seedling lots and the existence of reliable markers for taxa identification, it seems of prime necessity that the FRMs are delivered to the land owner with this information.