Whilst improving hygiene and sanitation behaviours is key to cost-effective and sustainable water, sanitation and hygiene interventions, measuring behaviour change remains a challenge. This study assessed the validity and reliability of pictorial 24-h recall (P24 hR), a novel method using unprompted recall of past activities through pictures, compared to structured observation for measuring handwashing with soap (HWWS) and safe child faeces disposal in rural Malawi. Data were collected from 88 individuals across 74 households in Chiradzulu district using both methods over a two-day period, with the recall period of the P24 hR corresponding to the period of structured observation completed the previous day. Results showed poor agreement between P24 hR and observations in detection of hygiene opportunities and behaviours. P24 hR under-reported handwashing opportunities when frequency was high and over-reported them when frequency was low. The 95% limits of agreement for handwashing opportunities estimated through Bland-Altman analysis (−7.62 to 4.89) were unacceptably wide given median 5 opportunities observed per participant. P24 hR also over-reported HWWS and safe child faeces disposal, and kappa statistics indicated agreement no better than by chance. Structured observation remains the better method for measuring hygiene behaviours as compared to the P24 hR method despite its known limitations, including potential reactivity bias.