Measuring safety performance is becoming increasingly important in many high-risk industries such as atomic power, the chemical industry, offshore oil production, air traffic control and construction. Much has been done to study the antecedents/factors that shape the safety culture and safety climate in these types of industries, but almost no research has been conducted into another high-risk industry – shipping. Based on the safety orientation model (SOM) and a review of items and scales used in surveys of safety climate and safety culture, a safety orientation scale (SOS) was developed and refined through the use of multivariate statistics. This study was conducted with a sample of seafarers sailing on Norwegian-owned vessels. A total of 2558 questionnaires were returned from 141 vessels and 16 shipping companies, giving a calculated response rate of 70%. The study showed that replicating previous studies on the sample of seafarers demonstrated a large degree of stability in scales and items across both industries and nations. Exploratory factor analysis (EFA) is the most commonly adopted approach to the development of scales of safety climate and culture scale reported in the literature, and factor retention seems to be the most important decision in EFA. In the present study several rules to determine the number of factors and items retained are applied, and a comparison is made of a short form of latent root criterion (SFLRC) and parallel criterion (PC). SFLRC is found to be the superior method for the present data set. Confirmatory factor analysis (CFA) did not support the preliminary SOM of 12 dimensions (13 dimensions, since one of the 12 dimensions was split during the CFA procedure); however, a re-specification of the model on the basis of the CFA for four different behavioural measures gave a simplified and well-defined model with seven factors and 22 items.