In Northern Laos, many farmers have switched within 15 years from a manual slash and burn system of long fallow and upland rice to a hybrid maize mono-cropping system with moto-mechanized plowing and chemical input use. This transition brought higher incomes to a majority of farmers, but sustainability of this system is henceforth challenged. This study identifies sustainability criteria that have to be considered to assess current cropping systems and alternatives. Our method considers 2 scales of analysis: 1. FARM level analysis identifies the diversity of farming systems and sustainability criteria from the farmer's perspective. It is based on 120 surveys in 6 villages (factorial analysis for mixed data and hierarchical clustering), focus groups discussions and card games (20 farmers in 4 villages). 2. FIELD level analysis identifies the factors explaining the variability of maize performances and impacts. It is based on surveys, 2-year monitoring of 38 plots, and a participatory game on tactical choices at cropping system scale (3 villages, groups of 20 farmers). At the farm level, we identified four farm types: the first type is composed of poor farmers, the second type is also poor farmers but with more land; farmers from the third type focus on rice production, whereas farmers from the fourth type focus on upland crops (maize, livestock, with higher total income). Sustainability criteria identified by farmers include: self-sufficiency in rice, farm transmissibility, farm income, diversity of activities to reduce risks, workload, and cash requirement for activities. Field level monitoring showed that poor performances of maize are explained by the difficulties to manage the first stages of the cropping cycle: poor soil tillage quality due to inadequate machinery leading to a heterogeneous stand with low density, erosion risks due to bare soil, strong weed pressure and consequently increasing use of herbicide. This diagnosis stage proved to be very useful as th