© 2019 Informa UK Limited, trading as Taylor & Francis Group.A large portion of the uncertainty involved in the design of high-pressure-high- temperature oil and gas pipelines installed subsea is associated with the variability in the shear strength of soil supporting the pipeline. This study investigates variability of shear strength in the top half-meter of the seabed at three deep water sites in the Gulf of Mexico. Nearly 400 undrained shear strength data from minivane tests conducted on soft clay from 38 box core samples are evaluated for the presence of spatial correlation (autocorrelation) structure. Findings show a high degree of scatter and a lack of spatial correlation in both undisturbed and remolded shear strengths. Classical linear regression modelling, which is applicable in the absence of autocorrelation, is used to determine (i) the mean strength, (ii) 95% confidence intervals for the mean strength and (iii) 95% prediction intervals for a single new observation for different depths. The linear regression approach implemented may be used to identify characteristic strength profiles corresponding to different levels of confidence. This approach produces strength profiles suitable for use in typical pipe-soil interaction analysis and is easily implemented. Although not new, to the author’s best knowledge, it hasn’t been used for seabed soils for pipeline design.