For the purpose of creating new forage crop suitable to warmer region of Japan, the authors intended to raise the interspecific and intergeneric hybrids using four main graminae harbacious plants, perennial ryegrass (Lolium perenne L.), Italian ryegrass (L. multiflorum Lam.), tall fescue (Festuca aruudinacea Schreb.) and meadow fescue (F. pratensis Huds.). For the first step, we studied on the crossability between these species, comparison of the morphological characteristics and cytological analysis of hybrid plants. The results obtained are as follows. 1. Intergeneric hybridization was not so hard, although the results were different in cross combinations, namely, we could get the hybrid plants more easily when Lolium sp. were used as maternal plants than reciprocal crosses. The interspecific hybridization was much easier than the intergeneric hybridization. But, back crosses to F_1 plants by parental species were very hard, and we got only three hybrid seeds which did not germinate. Furthermore, we could not get any seeds by sib-crossing between different F_1 plants or selfing in all hybrid plants tested. The percentage of seed setting by selfing differed in species and it was about 30% in tall fescue against 1.6% in perennial ryegrass. The results were compiled in Tables 1-5. 2. General appearance of hybrid plants resembled more to Festuca sp. than to Lolium sp., although dimentions of such characters as culm length, ear length, number of ears per plant, number of rachis per ear and number of nodes per ear were intermediate or resembled more to either parental species. Hybrid seeds were very light compared with parental seeds, and interspecific hybrid seeds were pretty heavier than intergeneric ones. In general, the degree of variation was large in hybrid populations. All hybrid plants delayed somewhat in the date of heading and flowering than parental species and showed long duration of flowering time. About 24% of the hybrid plants had completely degenerated anthers and the other showed very high pollen sterility ranging from 0.1 to 1.1%. The data on measured characters were presented in Figs. 1-3, Plate 4 and Tables 6-10. 3. As haploid chromosome number was 7 in Lolium sp. and 21 in Festuca sp., genus hybrid plants contained 28 chromosomes in somatic cells. In the meiotic division of the hybrid plants, it was not so easy to distinguish complete configurations, then we counted the number of univalent chromosomes for convenience. The frequency of univalent chromosome varied widely from 0 to 11 in either cell of the same anther or different plant. But, the average number of univalents did not differ so widely in cross combinations, ranging from 3.10 in L. mult.×F. prat. to 3.86 in L. mult.×F. arund. Such irregularities as chromosome bridges, laggards and fragments were also frequently observed in later stages of reduction division. From these analysis we supposed that the genome of Lolium sp. is semi-homologous to some genome of Fastuca sp. The results of cytological studies were compiled in Tables 11-14 and some microphotographs are presented in Plates 5 and 6. 4. We are now going to produce fertile amphidiploid plant by colchicine treatment or r-ray irradiation of hybrid seeds or F_1 plants.