Biophysical procedures have been used to determine the size and structure of the biologically active agent responsible for the transmission, through milk, of mouse mammary adenocarcinoma. Filtration of milk from RIII high-breast-cancer mice through gradocol membranes with decreasing pore sizes indicated that a minimum of activity passed through intermediate pore sizes (100 to 160 mmicro). Filtrates through smaller pores were significantly active. Milk treated with small doses of deuteron irradiation produced more tumors than the control, unirradiated milk; larger doses indicated a particle size much less than 100 mmicro. Free diffusion experiments indicated that the activity was associated with particles of two different sizes. Altogether the data denoted the presence of a large agent about 100 mmicro in diameter and a small agent 20 to 30 mmicro in diameter or possibly smaller. Furthermore, the presence in the milk of an inhibitor 40 to 60 mmicro is indicated by the results of all three approaches. The complex nature of the milk agent disclosed by the physical measurements agrees with the picture of one of the structures revealed by electron microscopy as well as with seemingly conflicting measurements reported in the literature. The large agent defined by these indirect methods corresponds to the whole particle seen in the electron microscope and the small agent corresponds to its internal core or nucleoid. It is suggested that the nucleoid is essentially a nucleic acid which may, in the absence of the "inhibitor," retain its activity after being stripped of its outer membrane or sac.