PACR of a project (8/84-12/91) to improve the institutional capacity of the Sri Lanka's National Water Supply and Drainage Board (NWSDB) and its delivery of health education and rural sanitation services. Early progress was hindered by an implementation approach that subjected NWSDB staff to radical organizational, attitudinal, and behavioral changes at a rate quicker than they were prepared to accept. in the beginning of 1986, several senior members of the NWSDB and the TA team were replaced and the project's approach was revised. While the new approach was slower and more incremental in nature, initial delays had already caused many of the project's activities to slip more than a year behind schedule. Thus the project was granted a 1-year extension to 8/91. a second extension, to 12/91, was made to accommodate NWSDB's purchase of a diesel generator. The project completed most of the planned organizational changes, trained staff, developed policies and procedures, and met objectives regarding NWSDB decentralization. However, due to civil turmoil in 1988-89, the construction of physical facilities, particularly regional offices, workshops, and laboratories, was delayed, and in some cases, terminated. Only 6 of 23 planned buildings were completed satisfactorily (3 more are scheduled for completion using NWSDB's own resources). Similarly, only 4 of 6 planned water supply subprojects were implemented. Fortunately, the problems with facilities construction did not affect decentralization, as other arrangements were made. Community participation promotion, health education, and latrine construction were accomplished in the 4 subproject areas. The project developed training curricula, constructed 5,400 adult latrines and 1,300 pre-school latrines, and trained some 370 volunteer village health workers in health education activities. The following were among the lessons learned. (1) a prescriptive design must be avoided at all costs in institution-building projects of this nature, since it places unnecessary burdens on both the institution and the TA contractor. (2) The key to successful institution-building effort in an engineering organization such as NWSDB is the selection of a TA consultant with a balanced mix of technical and organizational specialists. (3) Institution-building projects could benefit from the help of politically influential individuals. (4) All change strategies, new procedures, and other project interventions must be owned by the counterparts. From ownership comes acceptance and commitment, which in turn leads to sustainability. (5) Performance indicators must be developed, and they must be relevant and achievable. Performance review meetings must be held regularly and attended by top management. (6) The regular external monitoring carried out be the WASH project was highly successful in refocusing priorities, and was accepted by the both the contractor and NWSDB. The process included formal interviews and collaborative workshops attended by the TA team and chief NWSDB staff. (7) There should have been a more flexible approach to solving problems faced by contractors. Although the standard host country construction contract included a clause to cover price escalation, this was ordered deleted by USAID/SL. Then, implementation delays and subsequent inflation rendered contractors unable to meet costs, resulting in delays or default. Also, payment of contractors often took up to 6 months because four levels of approval were required (local firm, NWSDB regional office, NWSDB central office, and USAID). (8) The strong support of Sri Lankan and USAID officials at all levels was a key factor in the project's success.