Background Infections caused by multidrug-resistant organisms (MDROs) are a globally increasing threat among critically ill patients, especially those with underlying malignancies. We aimed to assess the prevalence and susceptibility patterns of MDROs among cancer patients in intensive care units (ICU), and their predictors. Methods Over 4 years, we retrospectively reviewed medical records of 497 malignancy patients in the ICU of a tertiary hospital in Alexandria, Egypt. The data for various factors, such as demographic characteristics, comorbidities, causative pathogen, and antimicrobial resistance (AMR), were collected and analyzed using univariate analysis. Logistic multivariate regression analysis was used to estimate the probability of developing MDROs among this population. Results A total of 748 isolates were obtained from 1249 specimens. Gram-negative bacteria detected (459) comprised 61.4% of all isolates, while only 75 (10%) were gram-positive, and 214 (28.6%) were fungal pathogens. The most frequently encountered isolate was Klebsiella pneumoniae (n = 183), of which 107 were carbapenem-resistant (CR) and 62 were extended-spectrum beta-lactamase (ESBL)-producing. This was followed by Escherichia coli (n = 136), of which 17 were CR and 100 were ESBL-producing strains, while 3 were resistant to quinolones. Acinetobacter baumannii came in third (n = 67), with 63 being CR. The overall susceptibility of gram-negative bacteria was recorded as highest to colistin (97.3%). The prevalence of methicillin-resistant Staphylococcus aureus (MRSA) and Enterococcal species among gram-positive bacteria were 54.6% and 33.3%, respectively, with no resistance reported to vancomycin or linezolid. Among the MDRO infection predictors were neutropenia, recent antibiotics use, and receiving chemotherapy. Neutropenia had the highest odds ratio (OR: 2.3, CI: 1.28–4.09), followed by recent antibiotics use (OR: 1.8, CI: 1.22–2.59). Conclusion Gram-negative bacilli were the most frequently reported MDROs, with resistance to higher generation cephalosporins and even carbapenems limiting antibiotic treatment options to older class antibiotics, such as colistin, with potential side effects, including nephrotoxicity. Estimating AMR probability using the prediction model of risk factors, such as neutropenia and previous antibiotics use, may be functional in the rapid identification of higher-risk patients.