Provided is a lossless/near-lossless image compression apparatus and method. The lossless/near-lossless image compression method may include calculating a predicted value by predicting a pixel of a current location based on adjacent pixels, calculating an error that is a difference between the predicted value and a pixel value, and modulating the error to output the modulated error, encoding the modulated error, and controlling a tolerance based on a slice unit to enable a number of encoded bits to be the same as a target number of encoded bits. The tolerance may be set to be gradually increased or to be decreased to prevent a rapid change in the tolerance.