In the open access environment, one of the challenges for utilities is that typical operating conditions tend to be much closer to security boundaries. Consequently, security levels for the transmission network must be accurately assessed and easily identified on-line by system operators;Security assessment through boundary visualization provides the operator with knowledge of system security levels in terms of easily monitorable pre-contingency operating parameters. The traditional boundary visualization approach results in a two-dimensional graph called a nomogram. However, an intensive labor involvement, inaccurate boundary representation, and little flexibility in integrating with the energy management system greatly restrict use of nomograms under competitive utility environment. Motivated by the new operating environment and based on the traditional nomogram development procedure, an automatic security boundary visualization methodology has been developed using neural networks with feature selection. This methodology provides a new security assessment tool for power system operations;The main steps for this methodology include data generation, feature selection, neural network training, and boundary visualization. In data generation, a systematic approach to data generation has been developed to generate high quality data. Several data analysis techniques have been used to analyze the data before neural network training. In feature selection, genetic algorithm based methods have been used to select the most predicative precontingency operating parameters. Following neural network training, a confidence interval calculation method to measure the neural network output reliability has been derived. Sensitivity analysis of the neural network output with respect to input parameters has also been derived. In boundary visualization, a composite security boundary visualization algorithm has been proposed to present accurate boundaries in two dimensional diagrams to operators for any type of security problem;This methodology has been applied to thermal overload, voltage instability problems for a sample system.