Replication or repetition is the means by which experimental error is estimated in replicated trials. In farm trials, it is sometimes not possible to replicate demonstration and test plots due to manageability and costs. In these situations, non-replicated observations and on-farm trials are used. In addition, nutrient response studies are often needed to calibrate the fertilizer requirement of a field. In this paper, we will define on-farm trials and observations, discuss the nature and underlying principles of non-replicated trials and observations, briefly discuss analysis methods, and provide relevant examples of non-replicated data analyses using SAS procedures.