FIVE PAKISTAN WATERCOURSE SYSTEMS, SELECTED FROM VARIOUS GEOGRAPHICAL AREAS OF THE INDUS BASIN, WERE EVALUATED DURING COMPLETE IRRIGATION TURN ROTATIONS WHILE OPERATING NORMALLY. THIS ALLOWED A QUANTIFICATION OF THE VARIOUS TYPES OF WATER LOSSES, INCLUDING TRANSIENT CONDITION LOSSES SUCH AS DEAD STORAGE, BANK WASH-OUTS, OUTLET LEAKAGE, AND HIGH INITIAL SEEPAGE INTO DRY CHANNEL BANKS. FLOW MEASUREMENT WAS MADE WITH CUTTHROAT FLUMES AND WATER VOLUMES WERE DETERMINED THROUGH INTEGRATION OF THE FLOW HYDROGRAPHS. TOTAL CONVEYANCE LOSSES RANGED FROM 38% TO 56% AND AVERAGED 45%. SIX TO EIGHT PERCENT OF THE INFLOW WAS CONSISTENTLY LOST TO TRANSIENT CONDITIONS OF WHICH ABOUT HALF WAS DEAD STORAGE. TRANSIENT LOSSES DEPENDED PRIMARILY ON THE LENGTH OF CHANNEL FILLED AND DRAINED. STEADY-STATE CONVEYANCE LOSS RATES WERE SIGNIFICANTLY HIGHER IN THE FARMERS' BRANCHES THAN IN THE MAIN CHANNELS, AND INCREASED RAPIDLY AS THE FLOW RATES INCREASED. SEEPAGE RATES INTO WATERCOURSE BANKS WERE MUCH HIGHER THAN INTAKE RATES INTO THE SURROUNDING FIELDS ON THREE WATERCOURSES, INDICATING A POTENTIAL FOR CONVEYANCE LOSS REDUCTION UTILIZING ONLY IMPROVED EARTHEN CHANNELS. APPLICATION EFFICIENCIES, MONITORED ON THREE OF THE STUDIED WATERCOURSE COMMAND AREAS VARIED WIDELY, AND AVERAGED 63%. FARMER WATER APPLICATION DID NOT CORRELATE WITH MEASURED ANTECEDENT SOIL MOISTURE DEFICIENCIES.