PACR for a project (9/82-9/91) to support community-based land and water management in upland forest, rain fed agricultural areas, and coastal zones in the Philippines. The project was divided into two cycles. Cycle I (1983-86) was mainly institution building for resource monitoring, policy analysis, and community-based resource management systems. Cycle II (1987-91) focused on field and farmer based oriented activities in the areas of agriculture, natural resources, research, and upland access. The project's research purpose was achieved. Several technologies were developed that support community-based land and water management, and mechanisms to support technology transfer (seed production, farmer-researcher-extensionist linkages, farmer-to-farmer approach, and applied communication) are being strengthened. The Department of Agriculture is capable of managing an expanded rain fed farming systems program as well as regional micro-level livelihood projects, while the Department of Natural Resources is being transformed from a regulatory agency to a developmental institution, capable of managing an expanded community-based agroforestry program and a contract reforestation program. In the area of natural resources, decentralization has been effected and NGOs/PVOs have actively participated in rehabilitation of degraded uplands. The access component did not meet targets, but did construct 73 km of rural roads and rehabilitate 64 km. From its experiences at more than 60 sites nationwide, the project has produced voluminous publications, videos, etc., including manuals and other tools for situation analysis, technology extension, and use of low-input technologies for upland and rain fed areas. The following are among the lessons learned. (1) Rolling design is not particularly appropriate to research programming, but could be effective in single agency projects having broad research areas and a single funding source. (2) Focusing research on specific areas does not effectively address needs of broader upland/rain fed areas. (3) Rapid rural appraisal, while useful, is expensive. (4) Requiring untrained extensionists to conduct technology adaptation trials is not a cost-efficient way to bridge the gap between research and extension. (5) Working within existing government systems increases the chance of a smooth transition upon project termination. (6) Sharing funding costs with implementing agencies promotes institutionalization. (7) Material support (e.g., seeds, agrochemicals, work animals) was important in gaining farmer cooperation. (8) Livelihood projects should emphasize a market-led approach and engage community groups instead of individual farmers. (9) Farmer-to-farmer training is effective, but care must be used in choosing farmer trainers. (10) Where development resources are limited, site selection is a sensitive issue that is influenced by political as well as technical realities. (11) Project staff should undergo a mandatory probationary period. (12) Technology development should be participative. (13) Opportunities for local capital building must be encouraged. (14) Appropriate uplands technologies include: multi-story cropping; livestock integration; communal reforestation; mangrove rehabilitation; and aquaculture. (15) For sustainable reforestation, encourage the growth of naturally occurring trees, which can then serve as "nurse trees" that provide better microenvironments for the more exotic reforestation species. (16) Labor-based, equipment-supported road construction creates jobs, doesn't sacrifice quality, and proved marginally less costly than equipment-based methods.