Embolized nitrogen appears in alveolar gas during clinical and experimental venous air embolism (VAE). Since early detection of VAE is believed to reduce morbidity and mortality, this study was done to compare the sensitivity of end-tidal nitrogen (ETN2) monitoring with other detection methods in current clinical use--precordial Doppler (PD), end-tidal CO2 (ETCO2), and pulmonary artery pressure (PAP). Ten mongrel dogs (10-17 kg) were anesthetized, placed in the supine position, immobilized, and ventilated (FIO2 1.0; PaCO2 35-40 mmHg). Anesthesia and muscle relaxation were maintained with constant infusions of thiamylal and pancuronium. Maintenance fluids were administered at 5 ml X kg-1 X h-1. Mean arterial pressure (MAP), PAP, and ETN2 and ETCO2 (Medspec II mass spectrometer) were displayed on a strip chart recorder. The dogs were divided into two equal groups and given either a step-wise sequence of 1-min air infusions (0.1-1.5 ml X kg-1 X min-1) or 5-s bolus air injections (0.25-1.0 ml X kg-1). Changes in PD sounds occurred in all animals at all air doses. Changes in cardiovascular variables and PaO2 were minimal. The threshold dose for ETCO2 and ETN2 to reach significance was 0.1 and 0.25 ml X kg-1, respectively, while PAP increases were significant at greater than 0.5 ml X kg-1 air doses. The time to maximum change (delta max) ETN2 was 30-90 s earlier than delta max ETCO2 (P less than 0.05) and 6-105 s earlier than delta max PAP. The delta max for all variables was dose related and statistically significant except for the smallest infusion VAE, where only ETCO2 was significantly changed.(ABSTRACT TRUNCATED AT 250 WORDS)