This paper presents a methodology to consider vague soil parameters required for earthwork optimisation, and to develop a genetic algorithm-based constrained curve-fitting technique required for highway vertical alignment process. The weighted ground line method is an earthwork optimisation methodology based on a hypothetical reference line and taking into account three soil properties to calculate realistic cut-fill volumes, namely swelling potential, compactibility percentage, and material appropriateness percentage. In this study, fuzzy rule-based inference methodology, which utilises previous experiences that can be expressed with linguistic terms, is employed to characterise swelling/shrinkage behaviour. In addition, material appropriateness concept is also adopted into developed optimisation methodology by a parametric algorithm using technical specifications and geotechnical data. Consequently, the genetic algorithm approach is employed for the determination of final grade line considering weighted ground elevations. The method involving an algorithm to consider the soil parameters as well as an evolutionary computation-based constrained curve-fitting technique produces outstanding geometric alignments.