Understanding plant resource investment and allocation is an important issue in ecology, since this can reflect the plant life-history strategies as a response to the changing environment. Many studies in the last decades have focused on plant resource allocation at different elevations, and most of them have determined that the plant resource allocation is strongly correlated with the plant size, representing a positive correlation between reproductive investment and plant size but a negative correlation between reproductive allocation and plant size. However, research has also shown that the reproductive allocation had no correlation with plant size. Significant elevation variation in the Qinghai-Tibetan Plateau provides the ecologists an ideal platform to reveal the plant resource allocations, since elevation variation can bring strong environment variances. However, previous studies did not address the variation in plant habitats, which could have directly affected the plant's resource allocation. In this study, we tested the reproductive allocation characters of Gentianopsis paludosa at three habitats located on the same elevation, on the east Qinghai-Tibetan Plateau: Kobresia sp. meadow (CD), Potentilla fruticosa shrub meadow (GC), and their junction area (JC). Centering the study on two levels, population and individual, we aimed to test the resource allocation characters of G. paludosa and reveal the relationship between the plant resource allocation and the plant habitats. Our results revealed the following: (1) The abiotic conditions were different among the three habitats, among which habitat CD displayed relatively better abiotic conditions compared with the others since the soil temperature and humidity, air temperature, and the net carbon absorptions were higher and optimal for plant growth. (2) At the population level, the individual size and reproductive allocation increased following the habitat variation of CD-JC-GC; the total flower numbers of G.paludosa did not differ significantly among the three habitats, whereas the bud numbers in the meadow were greater than those in the shrub meadow, and fruiting flower numbers in the meadow habitat were lower than those in the shrub meadow. (3) At the individual level, the reproductive biomass was positively correlated with individual size; a size threshold existed in G.paludosa and it declined with the habitat variation of CD-JC-GC; the reproductive allocation of G.paludosa had a negative correlation with the individual size, whereas the correlation coefficient declined following the habitat variation of CD-JC-GC; the flower number was positively correlated with individual size in all three habitats. These results suggest that reproductive allocation of G. paludosa is size-dependent, and that habitat variation can influence reproductive allocation and life-history strategies. The plant size and reproductive allocation characters were strongly related with their habitats, and reproductive allocation varied at both the population level and individual level. These variations were caused, primarily, in two ways: (1) the different natural conditions of the three habits could play an important role in determining the reproductive allocation characters; and (2) the genetic characters and the delayed-selfing breeding system were also correlated with the reproductive allocation. Although it was difficult to determine which of the two was dominant, the size-dependent reproductive allocation of G. paludosa in different habitats was undoubtedly a result of co-evolution between a plant and its habitat.