Guidelines for articulation of a framework for practical application of proportional-hazards models (PHMs) to professional survival analysis are provided. Focus is on data analysis fitting the PHMs with the semi-parametric methods of partial likelihood; this strategy is available in the BMDP2L and SAS PROC PHGLM computer programs. Areas in which advice and clarification are particularly apt include the definition of terms, the model and its assumptions, model-building, and interpretation and reporting of estimated model parameters. The presentation is based on a data-based example from a study of 10-year teacher survival patterns for a cohort of teachers who entered the profession in 1972 in Michigan. Survival analysis seeks to predict the duration during which a subject remains in a situation, in this case, the teaching profession. Discrete and continuous time analysis are examined; and the use of univariate, bivariate, and multivariate analysis are outlined. Choosing predictors and building a hierarchy of models, including interactions and predictors, and summarizing and comparing fitted models are discussed. Nine graphs and two tables are provided. (TJH)