A "random-hit" matrix model is proposed to account for the dynamic and steady state relationship between occupation of bovine renal medullary membrane receptors by [Lys8]vasopressin (LVP) and neurohypophyseal hormones (NHH) and the associated activation of membrane-bound adenylate cyclase. The model was developed by systematic introduction of specific rules concerning receptor coupling into a general structural model which consists of two square matrices of identical size, one composed of homogeneous R ("receptor") units, the second of homogeneous C ("cyclase") units. R units are either occupied (RO) or unoccupied (RU); C units are either active (CA) or inactive (CI). Hormone molecules are envisioned to "collide" with R units randomly; collision with RU leads to "binding", and occupation is maintained for a characteristic mean occupancy time, TO. In this structure, each R unit has an "interaction field" which consists of the "twin" unit in the "C" matrix, and the 4 nearest neighbor C units surrounding the twin. Occupation of an R unit leads to activation of all CI units in the interaction field of that R; CA units in the interaction field are refractory. Thus binding at a given R may "recruit" a variable number of inactive neighboring C units (5, 4, 3, 2, 1, or 0). The model requires that there be individual coupling delays between the moment of binding at a given R and subsequent activation of CI units (mean coupling delay (Td) approximately 10% To). Activation of C units persists as long as the "parent" R is occupied and is maintained for an additional short time interval (Tp) after RO reverts to RU, corresponding to hormone dissociation from receptor. The model accounts for the following previously demonstrated relations between LVP occupation of receptors and adenylate cyclase activation in bovine renal medullary membranes: 1) the shape of the nonlinear steady state relation between normalized (percentage maximal) receptor occupation (O) and cyclase activation (A), uniformly observed in different membrane preparations: 2) variable hormone concentration-dependent trajectories of approach to the final steady state A:O value (A:Oss) which may be either monophasic or biphasic; 3) the loss of intrinsic adenylate cyclase activity observed in bovine membranes for a series of NHH analogs with progressively diminishing affinity for receptors. The model represents an explicit theory of coupling where a successive series of temporal events are quantitatively related to each other and privide major constraints to any interpretation of the molecular organization of receptors and adenylate cyclase units in membranes. The model excludes a number of mechanistic proposals and suggests a new hypothesis for membrane coupling with features which may be generally applicable to other hormone-sensitive adenylate cyclase systems.