We present a method to simulate wells, i.e., 1-D stratigraphic profiles with attached physical properties but without spatial information, using a combination of geological knowledge and Monte Carlo statistics. The simulated data is intended to be used in seismic lateral prediction studies. Our algorithm simulates correlated stochastic variables one by one. There are two major advantages in this approach above the conventional way in which all correlated stochastic vectors are drawn simultaneously. The first advantage is that we can steer the algorithm with rules based on geological reasoning. The second advantage is that we can include hard constraints for each of the stochastic variables. If a simulated value does not satisfy these constraints, it can simply be drawn again. The input to the simulation algorithm consists of geological rules, probability density functions, correlations, and hard constraints for the stochastic variables. The variables are attached to the entities of a generic integration framework, which consists of acoustic-stratigraphic units organized at three scale levels. The simulation algorithm constructs individual wells by selecting entities from the framework. The order in which the entities occur, and the thickness of each entity, is determined by a combination of random draws and specified geological rules. Acoustic properties and optional user-defined physical properties are attached to the simulated layers by random draws. The acoustic properties are parameterized by top and bottom sonic and density values. The algorithm is capable of simulating acoustic hydrocarbon effects. The algorithm is demonstrated with a simulated example, describing the stratigraphic and physical variations in an oil field with a fluvial-deltaic labyrinth type reservoir.