How animals feel is both at the center of concerns about animal welfare and one of the most challenging questions for which to make strong scientific inferences. In this presentation, I will review various approaches used to investigate affective experiences in cattle, with a special focus on pain and fear (as much of the research is in this area). I will describe various methods available and argue that some of the most commonly used approaches suffer from important weaknesses. For example, provoked acute responses can be both highly variable and non-specific, meaning that both responses and non-responses are difficult to interpret. Stronger inferences can be drawn in cases where the functions of the responses are well understood, and with addition of appropriate control groups using drug treatments to target specific affective mechanism (e.g., analgesics in the case of pain). I will also review more elaborate approaches, including the use of approach-avoidance testing (that allows inferences about how much the affect matters to the animal), conditioned-place avoidance (that rely upon the memory of an experience), and drug-naming trials (that allow more specific inferences about affective states). Finally, I will discuss the potential of approaches that explore the longer-term impact of affective experiences, such a cognitive bias testing and anhedonia responses that may indicate mood changes and suggest more important welfare consequences.