Abstract: Abstract: As a light steel structure, plastic greenhouse is lighter in weight, shorter in service life, lower in cost, more sensitive to wind and snow load and less social effect than the industrial and civil building structures, such as masonry, reinforced concrete structures and even other light steel structures. Notable differences in features and characteristics between both structure designs are mainly as the follows. First, plastic greenhouse uses a flexible covering material (such as plastic film) which is far less intensive but more sensitive to the distribution and value of the wind and snow load. Its live load effect is much greater than the permanent load effect; therefore the design parameters (such as statistical parameters of resisting force and load effect ratio) of such structure are different from building structure. Second, greenhouse's parts can be removed, transported and reassembled, so the defects in bars and nodes inevitably occurs, which has a greater impact on the structure tension performance. Third, greenhouse is relatively short in service life, almost 5-10 years. Forth, due to fewer operators and flexible covering material, the social effects of failure for the plastic greenhouse are far less than the industrial and civil building structures. Therefore, according to the characteristics of the greenhouse above, by referring to the "Unified standard for the structure reliability design", the security level of plastic greenhouse can be classified as the third level, and the target reliability index is 2.70, equivalent to the failure probability of 3.5‰. At present, although the codes have been designed for building structure for a long time, there are no appropriate codes for such light structure in China, which leads to waste and unreasonable results inevitably. So it is especially important to improve the design method of plastic greenhouse. This paper studies the simple load combination of dead load and wind load. The dead load obeys the normal distribution, and by statistical hypothesis testing, the mean value of dead load is 1.06 times the standard value of dead load in the "Specification for the load of building structure" with the variance of 0.074 of the standard value of dead load. The wind load obeys the distribution of extreme value I. According to the data of 3 s extreme wind speed from 1981 to 2010, which were recorded by the domestic representative cities' meteorological stations, the formula of probability distribution of the maximum wind load through the design reference period is deduced as statistical basis. According to probabilistic limit state design method, the parameter values are determined which are target reliability index, load effect ratio, coefficient of variation of load, coefficient of variation of resistance, security level, and so on; the combination of permanent load and wind load is taken as the example, the partial factor for nominal value of loads is calculated to determine the partial coefficients for loads in combination of dead load and wind load. The statistics of constant load and variable load are the averages of random variables, and load partial coefficients are derived from characteristic value, so the average value should be converted to the characteristic value, and then the scope of the partial coefficients is iterated successively. The value of partial coefficient for dead load is between 1.07 and 1.11, and that for wind load is between 0.89 and 1.22. According to the scope of the partial coefficients, the theoretical values of the partial coefficients for structural parts under different load effect ratio are obtained, using the principle of least squares. The value of partial coefficient for dead load is 1.1 and that for wind load is 1.0. Finally, by comprehensive consideration, the suggested values are given, namely the value of partial coefficient for dead load and wind load is 1.0.