The mechanisms and physiological control of air-breathing were investigated in an extant halecomorph fish, the bowfin (Amia calva). Air flow during aerial ventilation was recorded by pneumotachography in undisturbed Amia calva at 20–24°C while aquatic and aerial gas concentrations were independently varied. Separation of aquatic and aerial gases was used in an attempt to determine whether Amia calva monitor and respond to changes in the external medium per se or to changes in dissolved gases within the body. Air flow measurements revealed two different types of ventilatory patterns: type I air-breaths were characterized by exhalation followed by inhalation; type II air-breaths, which have not been described previously in Amia calva, consisted of single inhalations with no expiratory phase. Expired volume (Vexp) for type I breaths ranged from 11.6+/−1.1 to 26.7+/− 2.9 ml kg-1 (95 % confidence interval; N=6) under normoxic conditions and was unaffected by changes in aquatic or aerial gases. Gas bladder volume (VB), determined in vitro, was 80 ml kg-1; the percentage of gas exchanged for type I breaths ranged from 14 to 33 % of VB in normoxia. Fish exposed to aquatic and aerial normoxia (PO2=19-21 kPa), or aerial hypercapnia (PCO2=4.9 kPa) in normoxic water, used both breath types with equal frequency. Aquatic or aerial hypoxia (PO2=6-7 kPa) significantly increased air-breathing frequency in four of eight fish and the ventilatory pattern changed to predominantly type I air-breaths (75–92 % of total breaths). When fish were exposed to 100 % O2 in the aerial phase while aquatic normoxia or hypoxia was maintained, air-breathing frequency either increased or did not change. Compared with normoxic controls, however, type II breaths were used almost exclusively (more than 98 % of total breaths). Type I breaths appear to be under feedback control from O2-sensitive chemoreceptors since they were stimulated by aquatic or aerial hypoxia and were nearly abolished by aerial hyperoxia. These results also indicate that Amia calva respond to changes in intravascular PO2; however, externally facing chemoreceptors that stimulate air-breathing in aquatic hypoxia cannot be discounted. Type II air- breaths, which occurred in aerial hyperoxia, despite aquatic hypoxia, appear to be stimulated by reductions of VB, suggesting that type II breaths are controlled by volume-sensitive gas bladder stretch receptors. Type II breaths are likely to have a buoyancy-regulating function.