Early season peaches and nectarines [Prunus persica (L.) Batsch] traditionally have relatively high market value, and increasing demand for high quality early season cultivars has made development of these cultivars a high priority for stone fruit breeders. For this reason, breeders are using more early ripening female parent materials in controlled hybridizations. Seeds from early season cultivars, however, are often immature when the fruit is ripe and will not successfully germinate with the normal procedure of stratifi cation for 2 to 3 months in a moist media at 5 °C. Immature seeds that do germinate often grow weakly and die or become stunted with rosetting symptoms. Many studies have investigated the diffi culty of germinating seed of early ripening stone fruit genotypes (Anderson et al., 2002; Connors, 1919; Kester and Hesse, 1955; Ramming, 1990; Tukey, 1933). Early studies found that a peach embryo develops along a sigmoid pattern during which the embryo is visible about 50 d after full bloom (DAFB), reaches full length (14 to 20 mm) about 80 DAFB, and fi nishes accumulating dry matter at about 140 DAFB (Chalmers and van den Ende, 1975; Tukey, 1933). To germinate and grow strong seedlings from immature seeds, breeders have developed embryo rescue techniques. Today embryo rescue protocols have been developed to an advanced degree and breeders routinely culture embryos as small as 2 mm in length (Byrne et al., 2000). However, while embryo rescue is a valuable and necessary tool for development of early ripening cultivars, use of traditional stratifi cation-germination is preferable when possible because stratifi cation-germination costs less and requires less labor and equipment. To avoid unnecessary use of embryo rescue, the breeder must be able to accurately assess the embryo maturity status of seed lots to be germinated. Most research to develop an index of embryo maturity has focused on the improvement of embryo rescue success. Embryo length (mm) and related measurements have been used for many years to determine the degree of embryo development of small embryos (Kester and Hesse, 1955; Ramming, 1990; Theobald and Hough, 1960). Embryo length is not an ideal developmental index because it does not consider the dry weight of the tissue, which is an important developmental indicator (Gage and Stutte, 1991). Furthermore, peach embryos upon reaching their full length at about 80 d of age are not yet viable and consequently embryo length has not been a useful tool for determining peach seed viability. Other tools for determining embryo maturity include FDP, embryo dry weight, and percent dry weight of the embryo. FDP is widely used by breeders as a general index to estimate embryo maturity but FDP is not always known, and even when known embryo maturity with respect to FDP may vary by year and by location. Much of this variability is related to average temperature during fruit development (Boonprakob et al., 1992; Braak, 1978; Gage and Stutte, 1991; Topp and Sherman, 1989). Generally, breeders have found that peach and nectarine embryos with 80 d FDP or less require embryo rescue for successful germination, and embryos with >100 d FDP can be germinated successfully with stratifi cation germination if they are stratifi ed before drying out. However, embryo maturity and germination success varies widely for genotypes with FDP in the 80to 100-d range. For these reasons breeders need a more accurate index of embryo maturity than FDP. Other possible developmental indexes include dry weight of the embryo or seed, and the ratio of embryo dry weight to embryo fresh weight, or percent dry weight of embryo (PDE). PDE has been considered a better indicator of embryo maturity than dry weight alone because seed size is inherently different for different cultivars (Pinto, 1992; Wareing and Phillips, 1981). Pinto (1992) determined that percent dry weight of the ovule (PDO = (ovule dry weight/ovule fresh weight) × 100) is correlated closely with PDE, and it has been shown that most of the dry weight increase of the ovule can be attributed to embryo growth (Chalmers and van den Ende, 1975). In breeding programs PDO would be more effi cient than PDE as an index of embryo maturity since the embryo does not need to be extracted from the seed coat in the case of PDO. Therefore, the objective of this study was to determine if PDO reliably estimates embryo maturity as measured by the percent seedling survival from stratifi cation germination. Fruit and germination data were recorded at College Station, Texas (lat. 30°35'N, long. 96° 21'S; elevation 96 m). The orchards are irrigated and peach trees received uniform cultural practices according to the recommendations for Texas peach orchards (Lyons et al., 1987). Days from full bloom to ripe (FDP) for each genotype was the main developmental index used. A peach tree was considered in full bloom when 60% to 80% of the fl owers were open, and the ripe date represents the date of fi rst commercial harvest (20% fruit fi rm ripe stage). In all genotypes, fruit was gathered when fully mature and stored in a cooler (5 °C ± 2 °C), usually 75% of unrotted seeds had emerging radicles (2 to 3 months). The seed was then planted into containers in a greenhouse maintained at 29 °C ± 3 °C (day) and 21 °C ± 3 °C (night). Because of the differences in ripening time and period required for germination, planting time extended from September through October. Surviving and rosetted seedlings for each seed lot were counted when >50% of the surviving normal (non rosetted) seedlings had grown at least 15 cm (2-3 months) and percent survival (PCTS) of seedlings was calculated: (number survived/number stratifi ed) × 100. In total, 170 seed lots, representing a range of 59 to 167 d FDP (mean = 97 d), 12% to 76% PDO (mean = 34%) and a range of chilling requirement (150 to 750 chilling units), were selected over 5 years (1994–98) to determine how well FDP and PDO predicted PCTS. This study was designed to represent the actual breeding program environment by using actual breeding germplasm over several years. Although most of the seed lots were from breeding materials, the following cultivars were also in the dataset: ‘Dixiland’, ‘EarliGrande’, ‘Fireprince’, ‘Flavorcrest’, ‘Flordaprince’, ‘Flordaking’, ‘Harvester’, ‘Hawthorne’, ‘June Gold’, ‘Juneprince’, ‘La Feliciana’, ‘Summerprince’, ‘TexKing’, ‘Texprince’, and ‘TexRoyal’. Data were taken for several years to minimize the possible effects that any particular year would have. Correlation analysis was done to quantify SHORT COMMUNICATIONS