Metabolizable energy (M.E.) values of 12 U.S. feedstuffs and 10 Korean feed ingredients for poultry were determined both by the total collection method and by the chromic oxide indicator method. It was found that M.E. values of most poultry feedstuffs can be measured accurately by either method. Limitation of feed intake to almost maintenance level(approximately 60% of ad libitum) did not increase or decrease the M.E. value of the feeds. An attempt was made to establish a prediction equation to estimate the M.E. values based on the apparent metabolizability of dry matter (D.M.) in the feedstuffs. The results indicated that linear relationships do exist between D. M. metabolizability and M.E. values of carbohydrate-rich feedstuffs (grains and their by-products) or protein-rich feed ingredients (oil seed meals and animal protein feeds) or lipid-rich feeds (fats and oils) as follows: The prediction equation for carbohydrate-rich feedstuffs was Y = 0.0947x - 3.498 ($r^2;=;0.99$, Sy.x = 0.015); for protein-rich feed ingredients. it was Y = 0.1234x - 4.898 ($r^2;=;0.99$, Sy.x = 0.027); and for lipid-rich feedstuffs it was Y = 0.0844x + 0.774 ($r^2;=;0.99$, Sy.x = 0.032). where x = metabolizability of dry matter of feeds in %, and Y=metabolizable energy values in kcal./g. The errors attached to these estimations were relatively small. Thus these prediction equations may be very useful for estimation of the M.E. values from D.M. apparent metaboiizability of feeds, especially in areas of the world where calorimetry is not possible.