Purpose Due to COVID-19 pandemic, the Italian population lived in quarantine from March to May 2020 (lockdown phase I). Restrictions impacted individuals' psychological health, especially in those with eating disorders (ED). Healthcare providers (HCPs) treating ED provided assistance by telemedicine and/or in walk-in clinics. We hypothesize that social restrictions represented a great stressor for ED patients and HCPs, negatively impacted their therapeutic alliance, and affected the frequency of dysfunctional behaviors. Methods This cross-sectional study consisted of an online survey investigating the experience of HCPs involved in ED treatment, with a specific focus on difficulties concerning the therapeutic efficacy. Questionnaire (n. 18 questions) was formulated ad hoc by our research team and sent by e-mail to Italian HCPs registered on online platforms. HCPs included ED experts specialized in psychology, nutrition or medicine. Data were collected during lockdown phase I and referred to patients with Anorexia Nervosa-(AN), Bulimia Nervosa (BN)-and Binge-Eating Disorder-(BED). Results One-hundred questionnaires were collected; 84 and 76 were included in our qualitative and quantitative analyses, respectively. Thirty-six% of HCPs felt their therapeutic intervention was unsuccessful, 37% complained compromised therapeutic alliance. Changes in frequency of compensatory behaviors (increased in 41% AN and 49,5% BN; reduced in 14,6% AN and 21,8% BN) and binge-eating episodes (increased in 53,3% BN and 30,5% BED; reduced in 30,7% BN and 24,7% BED) were experienced and ascribed to augmented patient's anxiety. Disorders switches and variation in dysfunctional conducts frequency were both significantly related to ED category (p < 0.05 for all). Concentration techniques were recognized as useful to offset such negative outcomes. Conclusion According to HCPs, social restrictions affected the frequency of dysfunctional behaviors in ED patients and the efficacy of their therapeutic intervention. Further long-term studies are needed to confirm our data in a larger sample size. Level IV Novel results from a cross-sectional study.