The objective of most pest management programs is to "control" the pest species. However, optimal control of local abundance and population growth may require different management strategies than optimal control of spatial spread. We use coupled demographic-dispersal models to address the relative importance of different management approaches to these two main control objectives for the invasive thistle Carduus nutans. The models are parameterized with data from thistle populations in the native (France) and invaded ranges (Australia and New Zealand). We assess a wide range of commonly used management strategies for their absolute and relative impacts on population growth and spread in both invaded-range scenarios. The projected population growth rate in New Zealand is more than twice that in Australia, while the spread rate is more than four times the Australian value. In general, spread and growth are both most strongly affected by the same life cycle transitions; however, in a few cases certain vital rates disproportionately affect either spread or growth. The transition that represents the contribution of large rosettes in one year to the number of large rosettes in the following year (the large rosette-large rosette transition) in Australia is dominated by reproduction (rather than survival) and hence is relatively more important to spread than to population growth. In New Zealand, the small rosette-small rosette transition is also predominantly dispersal-related. However, establishment of small plants from the seed bank contributes more to population growth than spread, as no dispersal is involved. The fine-resolution vital-rate-based modeling approach allows us to identify potentially novel optimal management strategies: approaches that reduce microsite availability show promise for reducing both population growth and spread, while strategies that affect dispersal parameters will affect spread. Additionally, the relative ranking of some biocontrol agents shifts depending on whether control of population growth or population spread is the desired outcome and therefore could alter which of the agents are preferred for release in a new area. The possibility of differences in ranked agent effectiveness has been predicted theoretically, but never before demonstrated using field data.