In the North Sea, it has been the practice to trench or bury all subsea pipelines below 16 " diameter to afford protection from the effects of fishing activities. However, the emergence of new guidelines for fishing interaction has resulted in pipelines below 16 " being left untrenched. These surface-laid pipelines are then susceptible to lateral buckling under operating conditions of temperature and pressure. Currently, one of two approaches is usually adopted for the analysis of lateral buckling. Analytical methods, based on closed form, solutions to the partial differential equation of flexure, are an extension of the work done on the buckling of rail tracks. These methods are straightforward to use but tend to suffer from several limitations, namely that of being restricted to linear material properties and small displacement theory, and generally giving conservative results. Numerical methods, such as the Finite Element Method (FEM) overcomes these difficulties and enable refinements such as variation in temperature profile, non-linear material properties and large displacement theory to be modelled. The majority of closed form solutions are independent of initial imperfections. In the FEM, however, it is necessary to instigate buckling by introducing an initial imperfection into the model. Both the pre-buckling and post-buckling behaviour is effected by the presence of initial imperfections. The variation in amplitude, wavelength and shape of the initial imperfections is addressed in terms of tendency to breakout. The cause of the initial imperfections is examined in the light of the pipelay barge motion and pipeline lay-tension. The traditional approach to lateral buckling, using analytical methods, can result in a conservative design that may require costly mitigation measures having to be taken. Adopting the FEM enables effects such as partial restraint and varying temperature profiles to be taken into account, and can result in significant cost savings for pipelines left untrenched. This can only be achieved if realistic initial imperfections can be defined. Incorporation of the effects of the initial imperfection into the design via a FE analysis is discussed.