Early detection of pests at a low population prevalence is a key pillar of surveillance for plant health. A novel pest is expected to undergo exponential growth when it first occurs in a host population. This allows for the use of mathematical rules of thumb for the number of samples required to detect presence before it reaches a target prevalence given the diagnostic performance of the tests used. Previous work assumes that pest presence is diagnosed by applying a single diagnostic test to each sample. However, diagnostic decisions are often made based on outcomes of multiple tests applied to the same sample in a consistent test program; for example, an assessment of symptoms followed by a PCR test applied to samples for which symptoms were observed. Each test can have different testing costs, as well as distinct and independent diagnostic performance and uncertainty values. A framework to optimize early detection surveys by minimizing overall costs for test programs that apply up to three diagnostic tests has been developed. The framework assesses the consequences of the test order and logical rule by which diagnoses are determined on costs across a range of pest prevalence. Explicit definitions of uncertainty in key parameters are incorporated to assess the consequences of uncertainty about their true value. Use of the framework is exemplified with two hypothetical case studies exploring the potential impact of selecting a suboptimal test program (based on Xylella fastidiosa) and investment in test improvements (based on lateral flow devices applied to Phytophthora pluvialis). [Formula: see text] Copyright © 2023 The Author(s). This is an open access article distributed under the CC BY-NC-ND 4.0 International license .