The accurate prediction of the Lightning Performance of Transmission Lines has been limited by many uncertainties. The range of possible variation of each parameter affecting line performance has been evaluated, A sensitivity analysis was performed to expose the effects of these uncertainties, using the Sargent-Darveniza backflashover computer program. The analysis revealed that backflashover rates are sensitive to the following parameters:-(i) the voltages due to the effect of charge in the stroke channel, (ii) the insulation sparkover voltages,(iii) the stroke current magnitude and waveshape(iv) the footing resistances ascribed to the tower base connections, and(v) the number of strokes to line. Investigation of these parameters then led to two significant improvements in the models used to evaluate them. (i) An improved physical model of the lightning stroke mechanism was used to re-evaluate the voltages due to the charge in the stroke channel. It is shown that this voltage component is much smaller than is predicted by the model of Wagner and Hileman. This conclusion is in accord with the reasonable success of prediction techniques based on current injection alone. (ii) A dynamic model to describe the non-linear surge current characteristics of footing resistances has been developed. This model accurately describes the surge behaviour of several earth connections in a variety of soils. It results in a reduction in the uncertainty from as much as 75% to about 5% under certain conditions. The use of the new model for studies of footing resistances at high lightning currents reveals that greater surge reduction factors than previously assumed can be expected. This study overcomes the previous difficulties in having to extrapolate values from meagre experimental results. An initial study of the relative attractive effects of earthed and unearthed lines suggests that the unearthed line has a smaller attractive radius. In this case, the number of strokes to line can be expected to be lower than for similar earthed lines. However, further field data is required to check this statement. A comprehensive Monte Carlo dynamic travelling wave computer program has been developed for calculating the lightning performance of unshielded distribution lines. The method has major advantages over previous methods. In addition to the concepts and data for backflashover calculations used by Sargent and Darveniza, the following improvements and new features necessary for unshielded line performance calculations are included: (i) Allowance for the presence of "weak-link" structures among normal structures. Such structures are known to have a disproportionate effect on line performance. (ii) Larger system simulated at 18 spans. This is necessary since surges can travel over considerable distances for preferred flashovers at weak-links.(iii) Allowance for corona wavefront distortion. (iv) Allowance for change in structure insulation characteristics following one or multiple flashovers across phases and/or to ground. (v) Dynamic representation of resistance of earth connections, using the analytical model proposed by the author. (vi) Adequate representation of lightning arresters. (vii) Multiple flashover assessment of transition from impulse flashover to power arc formation. This is a critical factor which has not been considered before. The adequacy of the developed method has been demonstrated by the good agreement of calculated outage rates and outage types, with those experienced on operating transmission lines. Arising from the analyses of these lines, corrective modifications for improved line performance have been recommended. Basic electrical requirements for good lightning performance of unshielded lines are established, and a study into several economical designs for achieving them is presented.