Social Desirability Bias (SDB), providing a favorable image of oneself in self-reports, is a persistent problem in survey research on sensitive topics. However, face-saving approaches appear promising in discouraging SDB. Specifically, face-saving survey item formulations aim to reduce the threat of judgment for norm-violations, thereby fostering a safer context for self-disclosure. This experimental study examined the use of face-saving to elicit more truthful responses to yes/no questions using a quota sample representative of the Dutch population. We included face-saving preambles that excused for or normalized sensitive behaviors (e.g., “It can easily happen that the limit for legal driving is exceeded when consuming alcohol”). We also included face-saving response options using degrees of truth (“occasionally”) or justifications for norm transgressions (e.g., “only when there was no other option”). There were three objectives. First, the design included a replication of a face-saving study regarding COVID-19, where we assessed the efficacy of a condition that combined a face-saving preamble with face-saving answer options (N = 303) against a control condition without any face-saving elements (N = 284). Second, we extended the replication’s design by including conditions to measure the effects of offering face-saving preambles (N = 281) and offering face-saving answer options in isolation (N = 283). Third, we investigated the effectiveness of face-saving using novel operationalizations on additional topics, including sustainability and responsible driving. The replication confirmed the effectiveness of the combined face-saving strategy. The extension that included all conditions (Ntotal = 1,151) revealed that offering face-saving answer options reduced SDB significantly, up to 23 percent. There was no main effect of including a preamble, nor was there an interaction effect of the two manipulations. These findings highlight the critical role that face-saving response options might play in improving the accuracy of self-reported data, while calling into question the usefulness of including face-saving preambles.