The stability and biological behavior of an in vitro system of doxorubicin (DXR) entrapped in NK911, polymer micelles, was examined and compared with those of DXR entrapped in Doxil, polyethylene-glycol-conjugated liposomes. The fluorescence of DXR inside micelles or liposomes in an aqueous solution is known to be strongly quenched by the outer shells of the micellar or liposomal formation. Thus, by measuring the fluorescence intensity of DXR released from NK911 or Doxil, we could determine the stability of the micellar or liposomal DXR formation. Furthermore, NK911 was found to be less stable than Doxil in saline solution. In drug distribution experiments using an in vitro solid tumor model, when spheroids formed from two human colonic cancer lines, HT-29 and WiDr, and a human stomach cancer line, MKN28, were exposed to NK911, DXR was distributed throughout the spheroids, including their center. On the other hand, when the spheroids were exposed to Doxil, DXR was distributed only to the surface of the spheroids. It has been suggested that Doxil can deliver DXR to a solid tumor more efficiently than NK911 via the EPR (enhanced permeability and retention) effect, because Doxil may be more stable in plasma than NK911. On the other hand, DXR packed in NK911 may be distributed by diffusion to cancer cells distant from the tumor vessel, because NK911 can leak out of the tumor vessel and may be able to release free DXR more easily than Doxil. It has been suggested that drug carrier systems such as liposomes and micelles should be selected appropriately bearing in mind the characteristics of the tumor vasculature and the tumor interstitium.