Evaluating the success of ecological restoration interventions in establishing self-sustaining development toward distant goals within project timescales is problematic. Trajectory analysis is a promising evaluation strategy to this end yet it has received little research attention and is uncommonly used. This thesis aims to identify indices predictable enough to be suitable for trajectory analysis, focusing plant assemblage structure. The primary objectives were to: a) accurately infer plant assemblage development gradients of primary successions in three different ecosystems of the South Island, New Zealand by means of sampling well aged chronosequences, and b) establish which indices had sufficiently strong and consistent response trajectories to all three inferred vegetation development gradients to be considered predictable. The vascular plant assemblages of at least five development stages in each of the three sites were sampled intensively using multiple fixed area plots. Ordination and stepwise regression established that age was highly correlated with the main floristic gradient and environmental variables were unimportant in explaining floristic variation. Data for index calculation consisted of plant species cover abundance and leaf area estimates as well as soil chemical properties. Development stage age estimates enabled index response trajectories to be constructed from stage mean values. Regression models were fitted to observed index trajectories for each site to test response strength and predictability. Comparisons of regression statistics and trajectories among the three sites for each index showed that the majority of indices had predictable responses to all sites; these were: soil pH and organic carbon, importance score, Simpson's species diversity, distance from the lognormal model of species relative abundance distribution, growth form diversity, taxonomic distinctness and DCA axis one. Together, these indices are suggested to be able to evaluate if development trajectories indicate progress towards three restoration goals via intermediate objectives. These goals are: 1) a persistent plant assemblage, 2) a plant assemblage with specific structural attributes and 3) a well functioning ecosystem. For trajectory analysis to effectively evaluate restoration success with these goals it is recommended that recovery gradients are long, monitoring periods are at least three decades and multiple indices are used that convey complementary information.