Rodents learn to fear a stimulus (e.g., a light) that signals the imminent arrival of an innate source of danger (typically an aversive foot shock). They also learn to fear a stimulus (e.g., a noise) that signals a learned source of danger (e.g., the already conditioned fear-eliciting light). Following Pavlov (1927), the former type of fear is termed first-order conditioned fear, because the stimulus is paired with an aversive unconditioned stimulus (US). The latter is termed second-order conditioned fear, because the stimulus is paired, not with a US, but with an already conditioned stimulus. There are both commonalities and differences in the neural substrates underlying these two forms of fear. Both require neuronal activity in the basolateral amygdala complex (BLA), including activation of NMDA receptors, for their encoding, and both require CaMK signalling, gene expression and DNA methylation for their consolidation. However, de novo protein synthesis is required for consolidation of first-order fear but not for consolidation of second-order fear.