Justice Jim Regnier delivered the opinion of the Court.

¶1 Annette Flink appeals from the Findings of Fact, Conclusions of Law and Judgment of the Workers' Compensation Court awarding her a 20 percent penalty on retroactive benefits, denying her request for attorney fees, and holding that the insurer properly calculated her weekly wage using a 40-hour work week. We reverse. There is one dispositive issue on appeal: Whether the Workers' Compensation Court erred in concluding that Flink's average weekly wage was properly calculated?

¶2 Flink was hired by Missoula Textile Services as a garment sorter and started work on May 19, 1998. On her first day of work, she suffered an injury to her left wrist. Flink worked approximately one and one-half hours the next day, May 20, 1998. She also worked for Missoula Textile Services between May 26 and June 11, 1998.

¶3 In September 1998 Flink hired counsel to assist her with her claim for workers' compensation benefits arising from her wrist injury. On September 23, 1998, Flink's counsel wrote American Alternative Insurance Co. (American), insurer for Missoula Textile Services, requesting that it initiate temporary total disability benefits retroactive to her last day of work. In response, American's claim adjuster began collecting medical and employment information on the claimant. In November 1998 Flink's treating orthopedic surgeon recommended surgery to treat her condition. American commenced paying temporary total disability benefits on December 14, 1998, retroactive to November 21, 1998, the day after Flink's treating orthopedic surgeon recommended surgery. On January 25, 1999, American paid Flink additional retroactive temporary total disability benefits for the period of July 14, 1998, to November 20, 1998. Two days later, it conceded liability for additional temporary total and temporary partial disability benefits for the period of June 12 through July 13, 1998.

¶4 On January 20, 1999, Flink filed a Petition for Hearing, claiming a dispute arose between the parties concerning the amount of benefits she was due. Pursuant to a pretrial order issued April 12, 1999, there were three issues to be determined by the court: (1) the number of hours Flink was hired to work; (2) whether American should be penalized for unreasonable delay; and (3) whether Flink was entitled to attorney fees and costs. The Workers' Compensation Court held a hearing on May 7, 1999. The court issued its Findings of Fact, Conclusions of Law and Judgment on May 25, 1999. The court concluded that Flink's wage was properly calculated using a 40-hour work week, granted Flink a 20 percent penalty for unreasonable delay on the retroactive benefits paid by American after she filed her petition, denied Flink's request for attorney fees, and awarded Flink's request for costs. Flink appeals.

¶5 We review findings of fact of the Workers' Compensation Court to determine whether they are supported by substantial credible evidence. See Matthews v. State Compensation Ins. Fund, 1999 MT 225, ¶ 5, 296 Mont. 76, ¶ 5, 985 P.2d 741, ¶ 5. We review the Workers' Compensation Court's conclusions of law to determine whether they are correct. Matthews, ¶ 5.

¶6 Whether the Workers' Compensation Court erred in concluding that Flink's average weekly wage was properly calculated?

¶7 The Workers' Compensation Court specifically found that it was more probable than not that Flink would have worked some overtime. The court then concluded that the amount of overtime she would have worked was speculative and based her compensation rate on a 40-hour week. The court stated that pursuant to Robertson v. Aero Power-Vac, Inc. (1995), 272 Mont. 85, 899 P.2d 1078, prospective overtime could not be used in calculating disability benefits unless an employer guaranteed the overtime or hired an employee to work a scheduled number of overtime hours. On this basis, the court concluded that Flink's wages were properly computed utilizing a 40-hour work week.

¶8 Flink asserts that an award of overtime wages does not depend on whether overtime was specifically scheduled or guaranteed. Flink contends that her rate computation should include overtime wages because she established that she would have worked overtime and she presented evidence of similarly situated employees who worked overtime. American counters that absent a clear indication that Flink was hired to work in excess of 40 hours per week, the trial court properly declined to speculate on prospective overtime. American also asserts that Flink failed to offer evidence sufficient for the trial court to accurately calculate overtime.

¶9 Flink's injury occurred while the 1997 version of the Workers' Compensation Act was in effect. Therefore, the 1997 version of the Act applies to her claims. See Madill v. State Compensation Ins. Fund (1997), 280 Mont. 450, 458, 930 P.2d 665, 670. An injured employee's wages for purposes of calculating workers' compensation benefits are determined pursuant to § 39-71-123, MCA. See Robertson, 272 Mont. at 87, 899 P.2d at 1080. Section 39-71-123(3)(a), MCA (1997), provides, in relevant part:

[I]f the term of employment . . . is less than four pay periods, the employee's wages are the hourly rate times the number of hours in a week for which the employee was hired to work.

¶10 The issue presented to the Workers' Compensation Court was the number of hours in a week for which Flink was hired to work. The court erroneously concluded that even though Flink proved she was hired to work overtime, pursuant to Robertson overtime could not be used in calculating wages unless an employer guarantees overtime or hires an employee to work a scheduled number of overtime hours. In Robertson, we did not conclude that overtime could only be used in calculating "the number of hours in a week for which the employee is hired to work" under § 39-71-123(3)(a), MCA, if it was "scheduled" or "guaranteed."

¶11 The issue in Robertson was the sufficiency of the evidence with regard to overtime-i.e., whether the Workers' Compensation Court's finding that Robertson was hired for 47 hours of work was clearly erroneous. On appeal, Robertson insisted that he was hired to work 5 to 6, 12-hour shifts and that therefore he was entitled to temporary total disability benefits based on a 72-hour work week. We affirmed the Workers' Compensation Court's determination that Robertson's benefits would be based on a 47-hour work week because he was not guaranteed a 72-hour week and because no employee who worked on the project worked for more than 47 hours. Robertson, 272 Mont. at 88, 899 P.2d at 1080.

¶12 Robertson does not stand for the proposition that overtime hours cannot be included in the hours per week for which an employee was hired to work unless those hours were guaranteed or scheduled. In fact, Robertson received a benefit based on a 47-hour work week despite the fact that he was neither guaranteed overtime nor scheduled a specific number of overtime hours.⁽¹⁾ Robertson was hired to work at a job that "could last five to six days" at shifts that could last "up to 12 hours." Robertson, 272 Mont. at 88, 899 P.2d at 1080 (emphasis added). The Worker's Compensation Court relied on an erroneous interpretation of Robertson in reaching its determination that Flink's benefits should be based on a 40-hour work week.

¶13 The Worker's Compensation Court's determination that Flink was hired to work only 40 hours a week is also not supported by substantial credible evidence. The court's findings with regard to the number of hours Flink was hired to work are inconsistent. The court found that Flink was hired on a 40-hour per week basis, but also found that had Flink continued her employment at Missoula Textile she would have worked overtime. If Flink proved that she would have worked overtime, then her rate should not have been computed on a 40-hour per week basis.

¶14 Moreover, the court also concluded that the number of overtime hours Missoula Textile hired Flink to work was "simply speculative." In every dispute concerning a prospective wage claim where the evidence clearly establishes that the employee would more likely than not work overtime, but the employer did not specifically guarantee or schedule overtime, the number of overtime hours the employee was "hired to work" is going to be speculative, at least in the sense that it will have to be based on circumstantial rather than direct evidence of the actual hours the employee worked. However, this does not preclude the consideration of overtime in calculating the claimant's rate under § 39-71-123(3)(a), MCA (1997). The Workers' Compensation Court demonstrated its facility to confront this issue in Robertson.

¶15 Lastly, we note that the record contains sufficient circumstantial evidence from which the Workers' Compensation Court could make a determination of the number of overtime hours Flink was hired to work. Dawn Steadele, the person who replaced Flink, had worked 21 hours of overtime in the last 31 weeks. Steadele did not request overtime. Rebekah Pierce, who requested overtime, testified that she performed similar, but expanded duties and averaged over 50 hours a week.

¶16 We reverse the Workers' Compensation Court's finding that Flink's disability benefits were properly calculated based on a 40-hour work week and remand for a redetermination of the number of hours she was hired to work. On remand the Workers' Compensation Court must necessarily address the further issues of attorney fees and penalty under §§ 39-71-2907 and -612, MCA (1997).

1. Interestingly, as in the instant case, the insurer in Robertson calculated the employee's rate based on a 40-hour week. However, unlike the instant case, the Workers' Compensation Court awarded the employee benefits on a 47-hour week even though the employee was not guaranteed or scheduled overtime.