A new approach to high resolution reservoir characterization is considered, based on the use of neural networks. These networks will be used to construct a transform which will accept 3D surface seismic depth converted data as input, and will produce estimates of six well logs as output, at each of the 3D seismic bin locations. Specifically, two distinct neural networks are to be trained, and then applied in succession to perform the necessary transform of low resolution surface seismic data into much higher resolution pseudo well log data, at each 3D seismic bin location. In order to aid in this process, crosswell seismic reflection data, which offers an intermediate stage of resolution, will be used as an integral part of the training of the two networks. The first neural network will be trained using interpolated surface seismic depth converted data and a set of computed attributes of these data, lying along each of two interwell seismic lines, as input. Target data for training will be crosswell seismic reflection data, together with the same set of attributes computed from those data. The second neural network will be trained using the crosswell seismic reflection data and their associated attributes as input, and will use a set of six existing well logs at each of three wells lying on the two interwell seismic lines as target data for training. If successful, the two neural networks, applied in succession, will produce estimates of six well logs at each 3D seismic bin location, using 3D surface seismic data as input. Results from this work are presented and discussed in some detail. Modifications were made to the original work plan, without significantly altering the approach outlined above, and some improvements were noted in the results. In the end, the results do not warrant application of this method for high resolution reservoir characterization in the reservoir being studied. However, the analysis of these results and issues associated with them may prove to be of value to anyone who might wish to consider a similar approach to high resolution reservoir characterization in the future.